HEALTHINF 2025 Abstracts

Full Papers Short Papers
Full Papers
Paper Nr: 27
Title:

Algorithmic Bias from the Perspectives of Healthcare Professionals
Authors:

Jennifer Xu and Tamara Babaian
Abstract: This paper focuses on algorithmic bias of machine learning and artificial intelligence applications in healthcare information systems. Based on the quantitative data and qualitative comments from a survey and interviews with healthcare professionals, who have different job roles (e.g., clinical vs. administrative), this study provides findings about the relationships between algorithmic bias, perceived fairness, and the intended acceptance and adoption of ML algorithms and algorithm generated outcomes. The results suggest that the opinions of healthcare professionals toward the causes of algorithmic bias, the criteria of algorithm assessment, the perceived fairness, and bias mitigation approaches may vary depending on their job roles, perspectives, tasks, and the algorithm characteristics. More research is needed to investigate algorithmic bias to ensure fairness and equality in healthcare.
Download
Paper Nr: 37
Title:

Mental Wellbeing at Sea: A Prototype to Collect Speech Data in Maritime Settings
Authors:

Pascal Hecker, Monica Gonzalez-Machorro, Hesam Sagha, Saumya Dudeja, Matthias Kahlau, Florian Eyben, Björn Schuller and Bert Arnrich
Abstract: The mental wellbeing of seafarers is particularly at risk due to isolation and demanding work conditions. Speech as a modality has proven to be well-suited for assessing mental health associated with mental wellbeing. In this work, we describe our deployment of a speech data collection platform in the noisy and isolated environment of an oil tanker and highlight the associated challenges and our learnings. We collected speech data consisting of 378 survey sessions from 25 seafarers over nine weeks. Our analysis shows that self-reported mental wellbeing measures were correlated with speech-derived features and we present initial modelling approaches. Furthermore, we demonstrate the effectiveness of audio-quality-based filtering and de-noising approaches in this uncontrolled environment. Our findings encourage a more fine-grained monitoring of mental wellbeing in the maritime setting and enable future research to develop targeted interventions to improve seafarers’ mental health.
Download
Paper Nr: 42
Title:

Virtual Dynamic Keyboard for Communication in Intensive Care
Authors:

Louisa Spahl and Andreas Schrader
Abstract: Effective communication is of great importance for intensive care patients in the weaning process to express their needs adequately. To support this process, the ACTIVATE patient application was developed, providing a selection of typically used texts via a novel interaction device BIRDY, intended to be used in bed. However, there are situations where patients would like to express more. Since the traditional layout of a static keyboard does not fit well with BIRDY gestures, we developed a virtual dynamic keyboard with letter prediction and minimal input gesture needs. We tested different text corpora and forecasting models, implemented a prototype based on the best candidate, and performed a preliminary user evaluation. The new virtual dynamic keyboard is shown to be superior compared to static layouts.
Download
Paper Nr: 57
Title:

Cross-Domain Transfer Learning for Domain Adaptation in Autism Spectrum Disorder Diagnosis
Authors:

Kush Gupta, Amir Aly and Emmanuel Ifecahor
Abstract: A cross-domain transfer learning approach is introduced to address the challenges of diagnosing individuals with Autism Spectrum Disorder (ASD) using small-scale fMRI datasets. Vision Transformer (ViT) and TinyViT models pre-trained on the ImageNet, were employed to transfer knowledge from the natural image domain to the brain imaging domain. The models were fine-tuned on ABIDE and CMI-HBN, using a teacher-student framework with knowledge distillation loss. Experimental results demonstrated that our method out-performed previous studies, ViT models, and CNN-based models. Our approach achieved competitive performance (F-1 score 78.72%) with a much smaller parameter size. This study highlights the effectiveness of cross-domain transfer learning in medical applications, particularly for scenarios with small datasets. It suggests that pre-trained models can be leveraged to improve diagnostic accuracy for neuro-developmental disorders such as ASD. The findings indicate that the features learned from natural images can be adapted to fMRI data using the proposed method, potentially providing a reliable and efficient approach to diagnosing autism.
Download
Paper Nr: 60
Title:

An Interpretable Machine Learning Model for Meningioma Grade Prediction
Authors:

Ali Golbaf, Damjan Veljanoski, Prutha Chawda, Swen Gaudl, C. Oliver Hanemann and Emmanuel Ifeachor
Abstract: Accurate preoperative prediction of meningioma grade is crucial for enhancing the clinical management of these tumours. In this study, we developed a non-invasive machine learning (ML) model to predict meningioma grade using clinical features and radiomics features from preoperative MRI scans, focusing on interpretability to improve clinical adoption of such models. A dataset of 94 patients from The Cancer Imaging Archive (TCIA) was analysed. Clinical features and radiomics features from T1-weighted contrast-enhanced (T1C) and T2-weighted Fluid Attenuated Inversion Recovery (T2 FLAIR) scans were utilised. Two feature subsets were constructed: one using radiomics features alone and the other combining clinical and radiomics features. Feature selection was performed using a modified Least Absolute Shrinkage and Selection Operator (LASSO) technique. Four ML models: Logistic Regression (LR), Support Vector Machine (SVM), Random Forest (RF), and Gradient Boosting (GB), were developed. SHapley Additive exPlanations (SHAP) was employed to address the blackbox nature of ML models by providing radiomics overall feature importance scores and model interpretation. Results using the clinical-radiomics subset showed that the SVM outperformed others (test AUC: 0.83), indicating its reliability for predicting meningioma grade. SHAP highlights discriminative radiomics features and their interaction with clinical features, thereby enhancing the clinical adoption of such models.
Download
Paper Nr: 64
Title:

Approximation of Inertial Measurement Unit Data to Time Series Kinematic Data Through Correlation Analysis and Machine Learning
Authors:

William Fröhlich, Rafael Bittencourt, Sandro Rigo, Rafael Baptista and César Marcon
Abstract: Accurate results are traditionally obtained in gait analysis using gold-standard methods such as motion capture with kinematic cameras and force platforms in biomechanics labs. However, these techniques are expensive, time-consuming, and require controlled environments, limiting their accessibility for more clinical and research applications. This study explores the potential of inertial measurement units as a cost-effective alternative. We focused on extracting features from Inertial Measurement Unit (IMU) data, such as acceleration and angular velocity, and derived metrics like speed and angular acceleration to approximate the accuracy of kinematic camera data. Following extensive preprocessing of inertial and kinematic datasets, we applied analytical methods, including Pearson correlation and cross-correlation, to identify significant relationships between the two data sources. We employed the most strongly correlated features to train Machine Learning models, Clustering techniques to assess the consistency and reliability of the results, and the Random Forest algorithm to train and evaluate the models’ capacity for time series prediction. Our findings suggest that certain aspects of IMU data strongly correlate with kinematic outcomes. This indicates that IMUs can replicate results traditionally obtained through more complex and costly methods under specific conditions.
Download
Paper Nr: 71
Title:

Privacy-Preserving Mortality Prediction in ICUs Using Federated Learning
Authors:

Pedro Vieira, Eva Maia and Isabel Praça
Abstract: Managing multiple patients in an Intensive Care Unit (ICU) can be extremely challenging. By predicting patient mortality, healthcare professionals can provide more efficient treatment and manage resources more effectively. This allows for more precise and useful interventions, potentially preventing fatalities. Although artificial intelligence (AI) is making significant advancements in this field, traditional Machine Learning (ML) continues to be the most widely used AI method, though it raises concerns about data security in collaborative environments. Since ensuring the safe handling of patients’ private data is crucial, Federated Learning (FL) has emerged as a viable alternative. Its intrinsic characteristics offer a valuable solution for training predictive models securely, as raw data does not need to be shared between participants. In this study, FL was used to develop models capable of predicting ICU patient mortality while protecting data privacy. Using data from the MIMIC-IV dataset, the most accurate model achieved an accuracy of 0.886, a recall of 0.817, and a specificity of 0.965, surpassing all the analyzed studies. A comparison between FL and traditional ML approaches revealed similar performance results. Moreover, three FL aggregation algorithms were evaluated, a less common focus in this area of research. Federated Averaging performed best with some classifiers, while delivering results comparable to FederAdagrad and FedAdam with others. In conclusion, the findings demonstrate that FL can be as effective as traditional ML for mortality prediction, with the added benefit of enhanced data privacy.
Download
Paper Nr: 107
Title:

Trends in Drug Prescriptions in the Outpatient Physician Sector in a German Federal State from 2014 to 2023 Using Morbidity Related Groups, Correlations and Partial Correlations
Authors:

Mareike Burmester, Timo Emcke, Vera Ries, Klaus-Peter Thiele, Bernhard van Treeck and Reinhard Schuster
Abstract: The pharmaceutical prescription data of all SHI-insured persons in a German federal state are analysed over a period of 10 years. With the help of the International ATC Code, each patient is assigned a Morbidity Related Group (MRG) as the active substance group with the highest costs per year. The leading MRG positions per age are compared between 2019 as the current year before the coronavirus pandemic and 2023 after the coronavirus pandemic. Between the ages of 23 and 31, treatment with antidepressants has come to the fore. Beta-lactam antibacterials and penicillins dominate in early childhood in both years and antithrombotics agents in old age. The correlations between age, polypharmacy and cost percentiles are examined in pairs or as a whole with correlations and partial correlations. All partial correlations of the three variables are greater than the correlations.
Download
Paper Nr: 118
Title:

Quality Clustering for Reducing the Search Space for Mobile Stroke Unit Allocation
Authors:

Muhammad Adil Abid, Johan Holmgren, Fabian Lorig and Jesper Petersson
Abstract: Mobile stroke units (MSUs), which are specialized ambulances equipped with a brain imaging device and staffed with trained healthcare personnel, have the potential to provide rapid on-site diagnosis and treatment for stroke patients. To maximize the efficiency of utilizing MSUs, it is crucial to strategically allocate these units. When solving the MSU allocation problem, the current methods search the whole search space when looking for the optimal solutions, which causes slow convergence. In the current paper, we propose the Quality Clustering for Reducing the Search Space (QCRSS) framework to reduce the search space by filtering out ambulance locations without negatively affecting the quality of the solution too much when solving the MSU allocation problem. By narrowing down the set of possible locations, the problem becomes more manageable, leading to faster convergence when solving the MSU problem. Extensive experiments under the multiple MSU settings show that the QCRSS is largely faster in convergence toward the optimal solution by reducing the search space by 5x, 11x, 26x, and 67x for two, three, four, and five MSUs, respectively. We illustrate the performance of the QCRSS through both qualitative and quantitative analyses.
Download
Paper Nr: 154
Title:

Leveraging Cross-Verification to Enhance Zero-Shot Prompting for Care Document Data Extraction
Authors:

Laura Steffny, Nanna Dahlem, Robert Becker and Dirk Werth
Abstract: Automating care documentation through artificial intelligence (AI), particularly using large language models (LLMs), has great potential to improve workflow and efficiency in healthcare applications. However, in clinical or care environments where errors can have serious consequences, ensuring the reliability and accuracy of LLM output is essential. Zero-shot prompting, an advanced technique that does not require task-specific training data, shows promising results for data extraction in domains where large, well-structured datasets are scarce. This paper investigates how cross-verification affects zero-shot prompting performance in extracting relevant care indicators from unbalanced nursing documentation. The extraction was evaluated for three indicators on a dataset of care documentation from 38 participants across two facilities. The results show cross-verification significantly improves extraction accuracy, particularly by reducing false positives. While term extraction alone achieved around 80% accuracy, at lower temperature settings (0.1) cross-verification increased accuracy to 96.74%. However, cross-verification also increased missed terms when no corresponding sentences were found, even though terms were in the ground truth. This study highlights the potential of cross-verification in care documentation and offers suggestions for further optimization, especially with unstructured text and unbalanced data.
Download
Paper Nr: 158
Title:

Integrating Gait and Clinical Data with Explainable Artificial Intelligence for Parkinson's Prediction: The EDAM System
Authors:

Nicoletta Balletti, Emanuela Guglielmi, Gennaro Laudato, Rocco Oliveto, Jonathan Simeone and Roberto Zinni
Abstract: Several machine learning (ML) approaches have been introduced for gait and posture analysis, recognized as crucial for early diagnosing neurological disorders, particularly Parkinson’s disease. However, these existing methods are often limited by their lack of integration with other clinical biomarkers and their inability to provide transparent, explainable predictions. To overcome these limitations, we introduce EDAM (Explainable Diagnosis Recommender), a system that leverages Explainable Artificial Intelligence (XAI) techniques to deliver both accurate predictions and clear, interpretable explanations of its diagnostic decisions. We evaluate the capabilities of EDAM in two main areas: distinguishing between healthy individuals and those with Parkinson’s disease, and classifying abnormal gait patterns that may indicate early-stage Parkinson’s disease. To ensure a comprehensive evaluation, we constructed one of the largest known dataset by merging and standardizing several existing datasets. This dataset includes 557 features and 7,303 labelled instances, covering a wide range of gait patterns and clinical features. Results show that EDAM achieves high accuracy in both tasks, demonstrating its potential for early detection of neurological disorders.
Download
Paper Nr: 159
Title:

Unveiling Breast Cancer Causes Through Knowledge Graph Analysis and BioBERT-Based Factuality Prediction
Authors:

Hasna El Haji, Nada Sbihi, Kaoutar El Handri, Adil Bahaj, Mohammed Elkasri, Amine Souadka and Mounir Ghogho
Abstract: Worldwide, millions of women are affected by breast cancer, with the impact significantly worsened in under-served regions. The profound effect of breast cancer on women’s health has driven research into its causes, with the aim of developing methods for the prevention, diagnosis, and treatment of the disease. The significant influx of research on this subject is overwhelming and makes manual exploration arduous, which motivates automated knowledge exploration approaches. Knowledge Graphs (KGs) are one of these approaches that attracted significant attention in the last few years for their ability to structure and present knowledge, making it easier to explore and analyze. Current KGs that include causes of breast cancer are deficient in contextual information, highlighting the uncertainty of these causes (facts). In this work, we present a method for extracting a sub-graph of breast cancer causes and fine-tuning BioBERT to evaluate the uncertainty of these causes. Our automated approach, which simulates human annotation, computes uncertainty scores based on textual factuality and assesses cause reliability using a Closeness Score. We also create a web-based application for easy explorationa.
Download
Paper Nr: 160
Title:

Optimizing Blood Transfusions and Predicting Shortages in Resource-Constrained Areas
Authors:

El Arbi Belfarsi, Sophie Brubaker and Maria Valero
Abstract: Our research addresses the critical challenge of managing blood transfusions and optimizing allocation in resource-constrained regions. We present heuristic matching algorithms for donor-patient and blood bank selection, alongside machine learning methods to analyze blood transfusion acceptance data and predict potential shortages. We developed simulations to optimize blood bank operations, progressing from random allocation to a system incorporating proximity-based selection, blood type compatibility, expiration prioritization, and rarity scores. Moving from blind matching to a heuristic-based approach yielded a 28.6% marginal improvement in blood request acceptance, while a multi-level heuristic matching resulted in a 47.6% improvement. For shortage prediction, we compared Long Short-Term Memory (LSTM) networks, Linear Regression, and AutoRegressive Integrated Moving Average (ARIMA) models, trained on 170 days of historical data. Linear Regression slightly outperformed others with a 1.40% average absolute percentage difference in predictions. Our solution leverages a Cassandra NoSQL database, integrating heuristic optimization and shortage prediction to proactively manage blood resources. This scalable approach, designed for resource-constrained environments, considers factors such as proximity, blood type compatibility, inventory expiration, and rarity. Future developments will incorporate real-world data and additional variables to improve prediction accuracy and optimization performance.
Download
Paper Nr: 162
Title:

Using Machine Learning to Assess the Impact of Harsh Violent Discipline on Children and Adolescents in Low- and Middle-Income Countries: A Comparative Analysis Focusing on Its Relationship with Disabilities
Authors:

Milena S. Barreira, Ariane C. B. da Silva, Hasheem Mannani and Cristiane N. Nobre
Abstract: Children’s exposure to violence has long been a social and cultural concern, manifesting in various forms across societies. According to UNICEF, approximately 300 million children worldwide, aged 2 to 4, experience regular violent discipline from caregivers, with around 250 million subjected to physical punishment. This study leverages data from the Multiple Indicator Cluster Survey to investigate the prevalence of severe violent discipline among children with and without disabilities in 54 low- and middle-income countries. Using machine learning algorithms, including Decision Tree, Random Forest, XGBoost, Support Vector Machine (SVM), and Neural Networks, the analysis revealed that SVM outperformed other models, achieving the highest precision, recall, and F1-score (with values of 78% and 80% for the violence and non-violence classes, respectively). The results highlighted an increase in severe disciplinary violence correlated with the presence of disabilities, particularly in contexts involving the domain of ‘controlling behavior’.
Download
Paper Nr: 171
Title:

HealthAIDE: Developing an Audit Framework for AI-Generated Online Health Information
Authors:

Tahir Hameed
Abstract: Online health information (OHI) encompasses a wide range of public-facing content, such as information on symptoms, diseases, medications, and treatments, while online medical information (OMI) involves more specialized and regulated content, including clinical trial data, surgical procedures, and medical research. OMI generation and dissemination is held to stringent standards for accuracy, transparency, and explainability, whereas OHI often requires information-seekers to independently evaluate credibility and relevance of the information. The rise of generative AI or large language models (LLMs) has exacerbated this disparity, as LLMs are primarily applied to public-domain OHI without sufficient safeguards, leaving users vulnerable to misinformation, bias, and non-transparent outputs. This paper presents a systematic literature survey on the usage of AI and LLMs in OHI, highlighting focus areas and critical gaps in developing a robust framework for auditing AI-generated health information. The proposed HealthAIDE Framework defines four key pillars for oversight: reliability and accuracy, trust and acceptance, security and safety, and equity and fairness. A short but systematic review of AI-driven health information literature reveals areas of stronger focus, such as accuracy and trust, and weaker focus areas, such as misuse prevention and transparency. Addressing these gaps through comprehensive audits will enable responsible evolution of AI-driven health information systems.
Download
Paper Nr: 175
Title:

Enhancing Diagnostic Accuracy of Drug-Resistant Tuberculosis on Chest X-Rays Using Data-Efficient Image Transformers
Authors:

Joan Jonathan Mnyambo, Amir Aly, Shang-Ming Zhou, Yinghui Wei, Stephen Mullin and Emmanuel Ifeachor
Abstract: Tuberculosis is an infectious disease with increasing fatalities around the world. The diagnosis of the disease is a major challenge to its control and management due to the lack of adequate diagnostic tools, contributing significantly to the prevalence of drug-resistant tuberculosis. Convolutional Neural Network (CNN) models have recently been developed to detect drug-resistant tuberculosis by analyzing chest radiograph images from the TB portal, but the classification results are low. This is because CNNs struggle to capture complex global and overlapping features in medical imaging, such as chest radiographs of drug-resistant tuberculosis. In contrast, transformers excel in these areas by utilizing self-attention mechanisms that detect inherent subtle and long-range dependencies across images. In this study, we used a pretrained data-efficient image transformer (DEiT) model to enhance the diagnosis of drug-resistant tuberculosis and differentiate it from drug-sensitive tuberculosis. The new model achieved an AUC of 80% in the detection of drug-resistant tuberculosis, an improvement of 13% in the AUC compared to current CNN models using data from the same source. The bootstrap significance test shows that the difference in AUCs is statistically significant. The results of the study can help healthcare providers improve drug-resistant tuberculosis diagnostic accuracy and treatment outcomes.
Download
Paper Nr: 196
Title:

CFC Annotator: A Cluster-Focused Combination Algorithm for Annotating Electronic Health Records by Referencing Interface Terminology
Authors:

Shuxin Zhou, Hao Liu, Pritam Sen, Yehoshua Perl and Mahshad Koohi H. Dehkordi
Abstract: In this paper, we present a novel algorithm designed to address the challenge of annotating electronic health record (EHR) text using an interface terminology dataset. Annotated text datasets are essential for the continued development of Large Language Models (LLMs). However, creating these datasets is labor-intensive and time-consuming, highlighting the urgent need for automated annotation methods. Our proposed method, the Cluster-Focused Combination (CFC) Algorithm, which stores intermediate results to minimize annotation loss from terminology-based annotators, such as BioPortal’s (mgrep), while achieving high coverage and significantly improving execution efficiency. We conduct a thorough evaluation of CFC on the benchmark dataset MIMIC-III, using the previously developed Cardiology Interface Terminology (CIT). Results show that CFC captured approximately 5,756 missed annotations from the baseline BioPortal (mgrep) while achieving a remarkable improvement in execution speed across different size of datasets. These findings demonstrate CFC’s scalability and robustness in processing large datasets, offering an efficient solution for EHR text annotation. This work contributes to the preparation of large, high-quality training datasets for Natural Language Processing (NLP) tasks in biomedical domains.
Download
Paper Nr: 242
Title:

Novel Approach to De-Identify Relational Healthcare Databases at Rest: A De-Identification of Key Data Approach
Authors:

Yazeed Ayasra, Mohammad Ababneh and Hazem Qattous
Abstract: Health information systems are widely used in the healthcare sector, and migration to cloud-based applications continues to be prominent in recent practices. Various legislations were issued by different countries to ensure the confidentiality of personal health information, introducing liabilities, and imposing penalties and fines to organizations in violation. This drives organizations to deploy significant investments in information security to safeguard various health information systems. The healthcare industry has experienced the second highest data breaches compared to other industries at 24.5% of the total data breaches in the United States between 2005 and 2023. Database layer vulnerabilities remain one of the most exploited resulting in attacks causing devastating confidentiality breaches for electronic personal health information (ePHI). The framework suggested in this work relied on de-identification using the health insurance portability and accountability act (HIPAA) safe harbor method of removing 18 identifying attributes from the data in its resting state. To achieve this, the work proposes 7 rules that allow the migration of health information system databases to the suggested framework structure to maintain a de-identified state of the database at rest. This is achieved through the segregation of identifying information in different tables based on their identification power and frequency of use while structuring them in a hierarchical manner where tables refer to the next or previous levels through encrypted foreign keys. The paper extends to successfully transform a typical EHR system database schema into a de-identified version of itself abiding to the 7 rules suggested by this work.
Download
Paper Nr: 252
Title:

Enhancing Fracture Aftercare Through a Human-Centered Mobile App Design
Authors:

Matthias Maszuhn, Felix Jansen, Frerk Müller-von Aschwege and Andreas Hein
Abstract: The rehabilitation process after fractures is crucial for achieving full recovery and maintaining patients’ quality of life, yet it faces growing obstacles due to demographic changes and healthcare resource shortages. This paper proposes a mobile app prototype for physiotherapy aftercare, integrating features like exercise assistance, load monitoring, and collaborative documentation to enhance patient support and accessibility. Employing a Human-Centered Design approach, requirements were gathered initially through a comprehensive literature review, followed by the creation of user personas. Final requirements were then refined through semi-structured interviews with physiotherapists and clinical staff. The prototype was subsequently evaluated in a user study with 16 participants using the Think-Aloud method and the User Experience Questionnaire (UEQ). Results indicated high user satisfaction with features such as education, exercise guidance, and progress tracking, though minor usability improvements were identified. By providing real-time feedback, clear progress tracking, and personalized guidance, the app aims to improve patient compliance with rehabilitation protocols, ensuring more consistent engagement throughout the recovery process. Future iterations will focus on expanding functionality and validating the solution across diverse demographics, emphasizing its potential to significantly improve rehabilitation outcomes.
Download
Paper Nr: 254
Title:

The Role of Digital Health Literacy and Socioeconomic Factors in Colorectal Cancer Screening: Machine Learning Analysis of HINTS Data
Authors:

Sujin Kim, Madhav Dahal, Avinash Bhakta and Jihye Bae
Abstract: While colorectal cancer (CRC) screening rates are on the rise, significant disparities persist, particularly among underserved populations, highlighting ongoing challenges in achieving equitable access to preventive care. This study utilizes machine learning models to analyze multi-year data from the Health Information National Trends Survey (HINTS), identifying critical factors influencing CRC screening adherence across three distinct time periods (2003–2008, 2011–2013, 2018–2020). Using Random Forest and Logistic Regression models, interpreted through Shapley Additive exPlanations values, we examine the impact of sociodemographic characteristics, digital health engagement, and digital literacy on CRC screening behaviors. Findings reveal that age, prior screening behavior, and digital literacy are key predictors; individuals with higher digital literacy, for example, exhibited a 22% higher likelihood of adhering to CRC screening guidelines. Age emerged as a dominant factor, with screening rates peaking at 43% in the 50–64 age group. These results suggest that interventions targeting digital health literacy and enhancing provider communication may effectively improve CRC screening rates among underserved populations. This study underscores the value of data-driven approaches in informing public health strategies to increase CRC screening adherence and reduce health disparities.
Download
Paper Nr: 267
Title:

Using Under-Represented Subgroup Fine Tuning to Improve Fairness for Disease Prediction
Authors:

Yanchen Wang, Rex Bone, Will Fleisher, Carole Roan Gresenz, Jean Mitchell, Wilbert van der Klaauw, Crystal Wang and Lisa Singh
Abstract: The role of artificial intelligence is growing in healthcare and disease prediction. Because of its potential impact and demographic disparities that have been identified in machine learning models for disease prediction, there are growing concerns about transparency, accountability and fairness of these predictive models. However, very little research has investigated methods for improving model fairness in disease prediction, particularly when the sensitive attribute is multivariate and when the distribution of sensitive attribute groups is highly skewed. In this work, we explore algorithmic fairness when predicting heart disease and Alzheimer’s Disease and Related Dementias (ADRD). We propose a fine tuning approach to improve model fairness that takes advantage of observations from the majority groups to build a pre-trained model and uses observations from each underrepresented subgroup to fine tune the pre-trained model, thereby incorporating additional specific knowledge about each subgroup. We find that our fine tuning approach performs better than other algorithmic fairness fixing methods across all subgroups even if the subgroup distribution is very imbalanced and some subgroups are very small. This is an important step toward understanding approaches for improving fairness for healthcare and disease prediction.
Download
Paper Nr: 270
Title:

Challenges of Generalizing Machine Learning Models in Healthcare
Authors:

Steven Kessler, Bastian Dewitz, Santhoshkumar Sundarara, Favio Salinas, Artur Lichtenberg, Falko Schmid and Hug Aubin
Abstract: Generalization problems are common in machine learning models, particularly in healthcare applications. This study addresses the issue of real-world generalization and its challenges by analyzing a specific use case: predicting patient readmissions using a Recurrent Neural Network (RNN). Although a previously developed RNN model achieved robust results on the Medical Information Mart for Intensive Care (MIMIC-III) dataset, it showed near-random predictive accuracy when applied to the local hospital’s data (Moazemi et al., 2022). We hypothesize that this discrepancy is due to patient demographics, clinical practices, data collection methods, and healthcare differences in infrastructure. By employing statistical methods and distance metrics for time series, we identified critical disparities in demographic and vital data between the MIMIC and hospital data. These findings highlight possible challenges in developing generalizable machine learning models in healthcare environments and the need to improve not just algorithmic solutions but also the process of measuring and collecting medical data.
Download
Paper Nr: 272
Title:

Early Diagnosis of Parkinson’s Disease via Pro-Saccadic Eye Movement Analysis: Multimodal Intermediate Fusion Framework
Authors:

Ji-Yun Han, Dae-Yong Cho, Dallah Yoo, Tae-Beom Ahn and Min-Koo Kang
Abstract: Early detection and timely treatment are essential for improving patient outcomes, but the lack of reliable biomarkers impedes early diagnosis of Parkinson’s Disease (PD). Consequently, eye movement abnormalities, known as early symptoms of PD, are gaining attention as crucial clues for early diagnosis. This study proposes a novel multimodal intermediate fusion framework for the early diagnosis of PD using eye-tracking data. The proposed framework improves the performance of classifying abnormal eye movement patterns in PD by integrating local features from time-series data and global features from encoded time-series images. Focusing on pro-saccade eye movements, this framework captures significant abnormalities like reduced peak saccadic velocity and multi-step saccades frequently observed in PD. The experimental results show a precision of 82% and a recall of 96% for PD, which demonstrates the effectiveness of the framework in minimizing missed diagnoses during early detection. In addition, this study highlights the potential of eye-tracking data as a biomarker for the early diagnosis of PD and predicts the advanced application of integrating wearable smart glasses for daily monitoring of neurodegenerative diseases.
Download
Paper Nr: 273
Title:

Affective Computing in Anxiety Disorders: A Rapid Literature Review of Emotion Recognition Applications
Authors:

Luigi A. Moretti, Miles Thompson, Paul Matthews, Michael Loizou and David Western
Abstract: Anxiety disorders (ADs) affect roughly one in ten people in the UK, and this number is expected to increase, intensifying the need for innovation. Digital technologies such as affective computing (AC, technology to detect human emotions) could foster a more patient-centric approach, enhancing therapy adherence and optimizing clinician-patient interactions. This paper reviews the literature relevant to the integration of affective computing in clinical pathways for ADs. A search was conducted on Google Scholar and PubMed using the keywords “affective computing” and subtypes of anxiety disorders. A total of 355 results were filtered to focus on peer-reviewed articles that specifically addressed emotion recognition in pathological anxiety as opposed to simply feeling anxious. Findings underscore prevalent studies focusing on post-traumatic stress disorder (PTSD) and the widespread use of valence and arousal for emotion quantification. Various approaches for both eliciting and detecting emotions are explored, offering technical and practical insights. Diverse applications, from monitoring treatment progression in behavioral therapies to assessing the efficiency of deep brain stimulation for intractable obsessive-compulsive disorder, highlight affective computing's versatility and promise. A significant advantage of digital technologies is their potential to capture longitudinal and contextualized data beyond clinical confines. Such assessments elucidate patients' daily challenges and triggers, enabling tailored interventions. The literature suggests that AC has the potential to support mental healthcare and improve patient outcomes. However, further evidence of its effective benefits is required, especially for ADs beyond PTSD, and further exploration of its implementation in clinical pathways is needed.
Download
Paper Nr: 296
Title:

Bag-Level Multiple Instance Learning for Acute Stress Detection from Video Data
Authors:

Nele Sophie Brügge, Alexandra Korda, Stefan Borgwardt, Christina Andreou, Giorgos Giannakakis and Heinz Handels
Abstract: Stress detection is a complex challenge with implications for health and well-being. It often relies on sensors recording biomarkers and biosignals, which can be uncomfortable and alter behaviour. Video-based facial feature analysis offers a noninvasive alternative. This study explores video-level stress detection using top-k Multiple Instance Learning applied to medical videos. The approach is motivated by the assumption that subjects partly show normal behaviour while performing stressful experimental tasks. Our contributions include a tailored temporal feature network and optimised data utilisation by additionally incorporating bottom-k snippets. Leave-five-subjects-out stress detection results of 95.46 % accuracy and 95.49 % F1 score demonstrate the potential of our approach, outperforming the baseline methods. Additionally, through multiple instance learning, it is possible to show which temporal video segments the network pays particular attention to.
Download
Paper Nr: 306
Title:

Extensive Conformance Testing and Validation of FHIR Data Exchange Variabilities
Authors:

Abderrazek Boufahja and Tanmay Verma
Abstract: The emergence of FHIR® standard during the last years was accompanied with the development of many FHIR® servers, some of them are commercials, and many are open-source projects, with a wide deployment in production. The FHIR® standard defines a complete RESTful API allowing access and sharing of clinical resources participating in dozens of healthcare workflows. The defined API comes with a complete list of variations in CRUD operations and in search queries. For instance, every search parameter comes with multiple searching flavours, making the implementation of the hundreds of search parameters complex, and the servers capability claims hard to verify by FHIR® clients, especially for those who use edge search capabilities. In this paper, we used a method to test exhaustively the large number of variabilities in the RESTful FHIR® API that can be implemented by a FHIR® server, by generating thousands of test scripts, using directly the formal description of the FHIR® standard. The method allows validating the different search variabilities and brings a deep view of the capabilities of the tested FHIR® servers. An implementation of the method was experimented, and the generated scripts were tested with multiple FHIR® servers. The testing of different FHIR® servers highlighted the conformance of most of them to the FHIR® standard, even if some discrepancies between the claims of some FHIR® servers and their current implementations were observed and analysed. We concluded the paper with an analysis of the search variabilities with commonly found behaviours and limitations. The overall work highlights the importance of a complete and strong testing strategy for a better integration and patient care.
Download
Paper Nr: 307
Title:

AI-Driven Early Mental Health Screening: Analyzing Selfies of Pregnant Women
Authors:

Gustavo A. Basílio, Thiago B. Pereira, Alessandro L. Koerich, Hermano Tavares, Ludmila Dias, Maria G. S. Teixeira, Rafael T. Sousa, Wilian H. Hisatugu, Amanda S. Mota, Anilton S. Garcia, Marco Aurélio K. Galletta and Thiago M. Paixão
Abstract: Major Depressive Disorder and anxiety disorders affect millions globally, contributing significantly to the burden of mental health issues. Early screening is crucial for effective intervention, as timely identification of mental health issues can significantly improve treatment outcomes. Artificial intelligence (AI) can be valuable for improving the screening of mental disorders, enabling early intervention and better treatment outcomes. AI-driven screening can leverage the analysis of multiple data sources, including facial features in digital images. However, existing methods often rely on controlled environments or specialized equipment, limiting their broad applicability. This study explores the potential of AI models for ubiquitous depression-anxiety screening given face-centric selfies. The investigation focuses on high-risk pregnant patients, a population that is particularly vulnerable to mental health issues. To cope with limited training data resulting from our clinical setup, pre-trained models were utilized in two different approaches: fine-tuning convolutional neural networks (CNNs) originally designed for facial expression recognition and employing vision-language models (VLMs) for zero-shot analysis of facial expressions. Experimental results indicate that the proposed VLM-based method significantly outperforms CNNs, achieving an accuracy of 77.6%. Although there is significant room for improvement, the results suggest that VLMs can be a promising approach for mental health screening.
Download
Paper Nr: 308
Title:

Holistic Cyber Threat Modeling for Machine Learning-Based Systems: A Case Study in Healthcare
Authors:

Janno Jaal and Hayretdin Bahsi
Abstract: Considering the immense pace in machine learning (ML) technology and related products, it may be difficult to imagine a software system, including healthcare systems, without any subsystem containing an ML model in the near future. However, ensuring the resiliency of these ML-based systems against cyber attacks is vital for more seamless and widespread technology usage. The secure-by-design principle, considering security from the early stages of development, is a cornerstone to achieving sufficient security at a reasonable cost. The realization of this principle starts with conducting threat modeling to understand the relevant security posture and identify cyber security requirements before system design. Although threat modeling of software systems is widely known, it is unclear how to apply it to software systems with machine learning models. Although adversarial machine learning is a widely studied research topic, it has yet to be thoroughly researched how adversarial and conventional cybersecurity attacks can be holistically considered to identify applicable cyber threats at the early stage of a software development life cycle. This paper adapts STRIDE, a widely-known threat modeling method, for the holistic cyber threat analysis of an ML-based healthcare system.
Download
Paper Nr: 323
Title:

Multimorbidity in Heart Failure Patients: Application of Machine Learning Algorithms to Predict Imminent Health Outcomes
Authors:

Jorge Cerejo, Rui Lopes Baeta, Simão Gonçalves, Bernardo Neves, Pedro Morais Sarmento, José Maria Moreira, Nuno André da Silva, Francisca Leite, Bruno Martins and Mário J. Silva
Abstract: As populations age and life expectancy increases, multimorbidity, which is the simultaneous presence of two or more chronic conditions, has become increasingly common, especially among older adults. Heart failure, a widespread and heterogeneous syndrome, has sparked research into multimorbidity to deepen our understanding of its pathophysiology and improve clinical management approaches. This paper offers a detailed characterization of a heart failure patient cohort, utilizing clinical data from a Portuguese tertiary hospital. Based on this characterization, we developed a clinical tool for identification of high-risk patients and prediction of imminent hospital admissions based on laboratory tests. Our models for predicting imminent hospitalization showed reasonable effectiveness (AUROC of 0.79 with lab test prescriptions and 0.72 with lab test results). These findings emphasize the significant predictive value of laboratory tests in the context of HF. Additionally, we investigated the explainability of our models using SHAP values, in collaboration with clinical experts, providing insights into factors influencing the models’ predictions. These results highlight the importance of secondary clinical data analysis assisting healthcare professionals in identifying patients at high risk of adverse events, and improving patient care and outcomes.
Download
Short Papers
Paper Nr: 22
Title:

Aspect-Level Sentiment Analysis of Filipino Tweets During the COVID-19 Pandemic
Authors:

John Paul S. Guzman, Charibeth K. Cheng, Jan Michael Alexandre C. Bernadas and Angelyn R. Lao
Abstract: During the COVID-19 pandemic, X (formerly known as Twitter) was teeming with rich discussions as people shared their experiences and concerns. Understanding the sentiments in these tweets could aid in gauging public reactions and enhancing public health communication. While some studies analyze public health sentiments, few specifically focus on aspect-level sentiments in the Global South. In this study, we examine tweets published in the Philippines during the COVID-19 pandemic and aspects relevant to the pandemic. The sentiment polarities of tweet-aspect pairs are annotated. We analyze these pairs to understand the sentiments expressed during this period. These insights can improve health communication in the Philippines by assessing public receptiveness to policies, monitoring events that influence sentiment, and identifying communication gaps. Notably, we observed disproportionately high amounts of negative sentiment toward the Sinopharm and Sinovac vaccines. This sentiment indicates distrust and racial bias against Chinese brands. Moreover, the consistent negative sentiment toward face shields over an extended period highlights shortcomings in health communication about their effectiveness.
Download
Paper Nr: 38
Title:

A Model-Checking Framework for Neuro-Degenerative Deficit Screening and Personalized Training
Authors:

Elisabetta De Maria and Christopher Leturc
Abstract: Serious games are established as an effective tool to screen cognitive deficits and assess diagnosis in patients affected by neuro-degenerative diseases such as Alzheimer or Parkinson. They are also known for their cognitive training benefits. According to the latest DSM-5 classification, we can discriminate mild Neuro-Cognitive Disorders (mild NCDs) and Major Neuro-Cognitive Disorders (Major NCDs). In this article, we consider three classes of patients: healthy, mild NCD, and Major NCD. For each class, we use Discrete Time Markov Chains to model the behaviour shown while playing serious games. Model checking techniques allow us to spot the difference between the expected and the observed behaviour. As a main contribution, we provide a new theoretical framework allowing us to evaluate how the confidence level of practitioners on the patient’s Alzheimer degree evolves after each game session, i.e., help to diagnose, and to set up an experimental protocol in which the levels of the proposed subsequent game sessions automatically depend on the patient behaviour observed in the previous sessions, i.e., help to train.
Download
Paper Nr: 41
Title:

Raising the Confidence of Mothers in Preterm Birth Care: Exploring the Secondary Role of the Internet
Authors:

Jamale S. El-Eid, Nabil Georges Badr, Salim M. Adib and Bernard Gerbaka
Abstract: Adequate family education and knowledge regarding basic preterm baby care is essential to enhance parents’ experience and alleviate the quality of life with preterm babies. Our study looks at the extent to which knowledge affects the confidence of new mothers. It explores other potential factors, sources of knowledge, and the role of technology and online content. The research model for our empirical investigation takes the foundations of the knowledge, attitudes, and practices (KAP) theory as the central survey framework of the theory of planned behaviour. The study results showed that NICU training has a significant impact on mothers' knowledge levels regarding the care of preterm babies after their discharge from the NICU. Findings revealed a prevalent reliance on unofficial online sources such as Google, social media, and other informal websites, rather than official resources like the WHO, CDC, or similar trusted platforms. Knowledge level emerged as a significant predictor of the dependent variable, maternal confidence, with a predictability score of 43.6%. This suggests that improved knowledge fosters greater confidence, particularly among first-time mothers who often rely on secondary internet sources to bridge their knowledge gaps and boost their confidence. These findings highlight opportunities for healthcare providers and health authorities to improve information generation and dissemination and foster support systems for parents.
Download
Paper Nr: 52
Title:

Formal Concept Analysis Applied to Characterize Longitudinal Associations Between Depressive and Anxiety Disorders and Somatization
Authors:

Diogo Miranda, Julio Neves, Luiz Zarate and Mark Song
Abstract: This study examines somatic syndromes as a significant public health challenge, highlighting the necessity of longitudinal sampling to comprehend the evolution of physical symptoms over time. It investigates the interplay between depressive and anxious symptoms and somatic symptoms related to disease. The research characterizes these symptoms within a diverse population in Isfahan, Iran, over a three-year period, utilizing Triadic Concept Analysis (TCA) as the primary analytical method to extract insights and establish correlations across time. The findings emphasize the importance of longitudinal methodologies in exploring patterns and rules associated with the symptoms under investigation. These insights enhance the understanding of the relationship between mental and physical health, offering valuable insights for clinical decision-making and treatment strategies.
Download
Paper Nr: 54
Title:

An Unsupervised Machine Learning Approach for Clustering Hip Arthroplasty Patients: Surgery Duration Differs Among Different Patient Groups
Authors:

Mohammad Chavosh Nejad, Rikke Vestergaard Matthiesen, Iskra Dukovska-Popovska and John Johansen
Abstract: Operating Rooms (ORs), as the largest source of revenue and costs in hospitals, face the challenge of growing demand while dealing with limited resources, emphasizing the need for operational efficiency. Duration of surgery (DOS), a key element in planning surgical resources, fluctuates and depends on many factors including patients’ characteristics. A better understanding of these factors and the way they affect DOS can help OR planners in achieving efficient resource allocation. To distinguish between patients from the DOS perspective, this paper proposes an unsupervised machine learning method that clusters patients into different groups by considering different clinical and operational features. Seven relevant factors were extracted from Aalborg University Hospital’s database for 1,847 patients undergoing hip arthroplasty. K-Prototype algorithm was utilized for developing various clustering models and their performance was assessed by three popular metrics. Among the different developed models, the one with 7 clusters achieved the highest performance. One-way ANOVA analysis illustrated that DOS means are significantly different among different clusters (F-statistic=11.77, P-Value=5.45e-13). Inter-cluster differences were analyzed by Turkey’s Honest Significant Difference (HSD) test. Besides, evaluating features’ importance showed that Age, BMI, and surgery type are the most contributing factors in clustering patients.
Download
Paper Nr: 63
Title:

AI Models for Ultrasound Image Similarity Search: A Performance Evaluation
Authors:

Petra Takacs, Richard Zsamboki, Elod Kiss and Ferdinand Dhombres
Abstract: Querying similar images from a database to a reference image is an important task with multiple possible use-cases in healthcare industry, including improving labelling processes, and enhancing diagnostic support to medical professionals. The aim of this work is to measure the performance of different artificial neural networks, comparing their ability to identify clinically relevant similar images based on their generated feature sets. To measure the clinical relevance, metrics using expert labels of organs and diagnoses on the images were calculated, and image similarity was further confirmed by pixel metrics. Images with organ and diagnosis labels were selected from a dataset of early-stage pregnancy and 2nd -3rd trimester pregnancy ultrasound images respectively for the measurements. The networks were chosen from state-of-the-art foundational models trained on natural images, DINO and DINOv2, SAM2, and DreamSim. The best performing model based on our experiments is DreamSim for organ matches, and DINO for diagnosis matches. A simple ResNet trained on the mentioned early pregnancy dataset for organ classification was also added to the selection. ResNet performs best for early pregnancy organ matches, therefore finetuning a robust encoder on our own dataset is a promising future step to further enhance medically relevant similar image search.
Download
Paper Nr: 73
Title:

Online Communities for Promoting Physical Activity: A Scoping Review of Use, Characteristics and Research Gaps
Authors:

Jennifer Hachiya
Abstract: The objective of this scoping review was to identify, characterise, and synthesise existing literature on the use of online communities (OC) to promote physical activity (PA) and identify gaps to direct future research. Systematic searches were conducted in Science Direct, PubMed, Scopus, and Institute of Electrical and Electron-ics Engineers Xplore for studies published up to August 2020. The search terms included a combination of the following keywords: physical activity, sedentary, exercise, health, sport, brand, and online community. No limits were used. Studies were included if they encompassed a full publication containing enough details on characteristics and described any feature primarily aiming at PA promotion. A total of 21 different OC were found in the total of 25 selected studies. Of those studies, all reported on at least one behaviour change technique, 68.2% (n=15) used websites to support the OC, 36% (n=9) reported on strategies to keep users engaged, 16% (n=4) comprised information related to the design process, and 16% (n=4) reported on OC effectiveness. Existing reports do not provide evident detailed information on the design process or user engagement strategies related to OC, and only a few studies assess its effectiveness in improving PA. Further research is needed.
Download
Paper Nr: 74
Title:

Analysing Digital Platforms and Online Communities for Promoting Physical Activity
Authors:

Jennifer Hachiya
Abstract: The objective of this study was to analyse existing digital platform (DP) characteristics of online communities (OC) to promote physical activity (PA). Previously DP identified in our previous scoping review were matched against our inclusion criteria. DP were included if mainly used to promote PA and were free of access. In addition to the general attributes of each DP, data was retrieved on user engagement strategies, BCT, and platform credibility. A total of 50 DP were found in our Google search. Fourteen OC from the Google search and 3 OC from our previous scoping review (n=17) were included in this study. Most DP (13; 64.70%) use an activity tracker—either external or internal—to support users on PA self-monitoring, almost all DP (16; 94.12%) included GPS connectivity features, and about half of selected DP (9; 52.94%) had a forum for community interaction. We found references to 26 (92.86%) of the 28 strategies used for analysis. While research on OC to promote PA and DP characteristics has been growing, existing DP does not provide detailed information on its attributes, nor comprehensive, specific data on engagement strategies and BCT.
Download
Paper Nr: 78
Title:

Quantifying the Role of Active Listening and Reassurance in Virtual Health Coach Interactions
Authors:

Ghulam Hussain, Brian Keegan and Robert Ross
Abstract: Conversational Agents have the potential to support healthcare through coaching exercise routines, but are still lacking in demonstrating authentic social behaviours to support engagement. To this end, we present a series of experiments that we conducted in order to investigate how automated health care coaches can be more effective when their interaction style is tailored to demonstrate qualities associated with a good bedside manner, namely active listening and reassurance. To test this, we first developed a dataset of 135 dialogue excerpts from three distinct sources, i.e., original, handcrafted and LLMs, the latter two of which were tuned to demonstrate specific types of comforting or reassuring language. Using this dataset, we conducted a study to validate whether users perceive different levels of active listening and reassurance across sources. The results of the study indicate that users can distinctly perceive the varying levels of stimuli across the three different data sources and that LLMs in particular clearly demonstrate these properties. In an accompanying analysis, the results showed that there is no notable influence of participant personality on perception, which we argue reduces the barrier to successful system deployment.
Download
Paper Nr: 82
Title:

Assessing Signal Noise Effects on Machine Learning Models for ECG-Based Cardiac Diagnosis
Authors:

Emanuela Guglielmi, Davide Donato Russo, Pasquale Trinchese, Gennaro Laudato, Simone Scalabrino, Gianluca Testa and Rocco Oliveto
Abstract: The Internet of Medical Things (IoMT) plays a vital role in healthcare by enhancing preventive careand chronic disease management through continuous monitoring using smart sensors and wearable devices. However, the reliability of IoMT systems can be compromised by noise in the acquired vital signals, which can negatively impact the accuracy of Machine Learning (ML) models used for anomaly detection. This study evaluates the impact of various disturbances on the performance of ML models in predicting cardiac conditions, with a focus on assessing the reliability and effectiveness of these systems in real-world applications. We investigated the effects of three types of noise—baseline wander, muscle artifact noise, and electrode motion artifact—on the performance of two advanced ML models designed to predict cardiac conditions, specifically atrial fibrillation (AF) and ventricular tachycardia (VT). Our analysis centered on how different noise intensities (i.e., the “loudness” of the noise) and durations (i.e., the length of time the noise persists) impacted the classification performance of these models. The VT detection model showed robust performance, with minimal impact even under intense and prolonged noise conditions. In contrast, AF detection was affected by all types of noise, with classification accuracy decreasing by up to ∼59% in the most challenging scenarios.
Download
Paper Nr: 83
Title:

VIRTUAL-PHYSIO: A Virtual Assistant for Home Physiotherapy Rehabilitation
Authors:

Nicoletta Balletti, Antonella Cascitelli, Patrizia Gabrieli, Emanuela Guglielmi, Gennaro Laudato, Aldo Lazich, Marco Notarantonio, Rocco Oliveto, Stefano Ricciardi, Simone Scalabrino and Jonathan Simeone
Abstract: Mobility impairments reduce the ability of patients to complete daily activities. Physio-therapeutic exercises help patients address such limitations. Correctly executing these exercises is crucial, often requiring a physiotherapist’s guidance. To address this need, combining advanced sensors with artificial intelligence offers a promising solution for home rehabilitation, enabling remote monitoring and reducing stress. In this paper, we introduce VIRTUAL-PHYSIO, a virtual assistant for remote rehabilitation integrated into a home-deployable low-cost physiotherapy monitoring system 2VITA-B PHYSICAL. VIRTUAL-PHYSIO provides real-time feedback during rehabilitation exercises and evaluates entire sessions, allowing physiotherapists to focus on critical cases. We experimented with VIRTUAL-PHYSIO on 51 individuals whose performances were also evaluated by a physiotherapist as a reference. The results (i) highlight good patient acceptability for the virtual assistant, and (ii) show that the proposed machine learning approach can effectively perform an automated evaluation of rehabilitative movements.
Download
Paper Nr: 96
Title:

The Use of Formal Concept Analysis for Characterizing the Behavior of the Residents of Bangladesh Regarding the COVID-19 Pandemic
Authors:

Mateus Sobreira, Martha Dias, Luiz Zarate and Mark Song
Abstract: COVID-19 reported its first case on December 31, 2019, an illness rapidly spreading, becoming a global pandemic. Due to transmission through contact with respiratory droplets from infected individuals, governments implemented various preventive measures to minimize the spread of the disease. This work utilized Formal Concept Analysis (FCA) to assess the behavior of the inhabitants of Bangladesh regarding COVID-19, based on responses to a public questionnaire conducted virtually in 64 districts from April 1 to 10, 2020. A preprocessing stage was performed on the data to fit them into the format of a formal context with objects, attributes, and the relationships among them. From the resulting general formal context, it was possible to subdivide it according to the respondent gender, enabling an analysis of the behaviors of the men and women. Based on these formal contexts, association rules containing the most predominant relationships in the database were filtered using thresholds of 40% support and 80% confidence.
Download
Paper Nr: 97
Title:

Usability Evaluation of a Chatbot for Fitness and Health Recommendations Among Seniors in Assisted Healthcare
Authors:

William Philipp, Ali Gölge, Andreas Hein and Sebastian Fudickar
Abstract: This study explores the acceptance of seniors for a chatbot designed to support in maintaining activity levels and quality of life in an assisted healthcare setting. Building on findings from the TUMAL study, which developed a self-assessment tool for physical functioning, a proof-of-concept chatbot was created as an Android app. The chatbot enables users to view their health data, inquire about activity levels, and receive recommendations based on their results. A study involving 12 seniors (aged 75+) was conducted to evaluate the chatbot's usability and the participants' attitudes toward its recommendations. The System Usability Scale (SUS) revealed a suboptimal usability score of 66.3, with wide-ranging results indicating varying user experiences. While fitness-related recommendations were positively received, health-related advice prompted mostly negative feedback. Despite these challenges, the data querying functionality was considered useful, demonstrating a degree of acceptance among the senior user group. The study suggests that the participants' technical proficiency may have influenced their overall usability ratings.
Download
Paper Nr: 101
Title:

Optimization of a Deep-Learning-Based Cough Detector Using eXplainable Artificial Intelligence for Implementation on Mobile Devices
Authors:

P. Amado-Caballero, I Varona-Peña, B. Gutiérrez-García, J. M. Aguiar-Pérez, M. Rodriguez-Cayetano, J. Gomez-Gil, J. R. Garmendia-Leiza and P. Casaseca-De-la-higuera
Abstract: Respiratory diseases, including COPD and cancer, are among the leading causes of mortality worldwide, often resulting in prolonged dependency and impairment. Telemedicine offers immense potential for managing respiratory diseases, but its effectiveness is hindered by the lack of reliable objective measures for symptoms. Recent advances in deep learning have significantly enhanced the detection and analysis of coughing episodes, a key symptom of respiratory conditions, by leveraging audio signals and pattern recognition techniques. This paper introduces an efficient cough detection system tailored for real-time monitoring on low-end computational devices, such as smartphones. By integrating Explainable Artificial Intelligence (XAI), we identify salient regions in audio spectrograms that are crucial for cough detection, enabling the design of an optimized Convolutional Neural Network (CNN). The optimized CNN maintains high detection performance while significantly reducing computation time and memory usage.
Download
Paper Nr: 110
Title:

Federated Learning in Multi-Center, Personalized Healthcare for COPD and Comorbidities: The RE-SAMPLE Platform
Authors:

Jakob Lehmann, Gesa Wimberg, Serge Autexier, Alberto Acebes, Christos Kalloniatis, Costas Lamprinoudakis, Thrasyvoulos Giannakopoulos, Andreas Menegatos, Agni Delvinioti, Giulio Pagliari, Nicoletta di Giorgi, Jarno Raid, Danae Lekka, Aristodemos Pnevmatikakis, Sofoklis Kyriazakos, Konstantina Kostopoulou and Monique Tabak
Abstract: Federated learning is becoming more and more popular, also in healthcare applications. The platform, developed within a multidisciplinary consortium, is enabling privacy-preserving training of machine learning models generating predictions for patients with chronic obstructive pulmonary disease and comorbidities. Moreover, data synchronization and monitoring is made possible using the HL7 FHIR standard. The platform provides two front ends; a patient facing smartphone app and a healthcare professional facing dashboard that is used inside three different hospitals in Italy, Estonia and the Netherlands. The overall architecture and implementation into practice is shown in this paper.
Download
Paper Nr: 122
Title:

Can We Trust Explanation! Evaluation of Model-Agnostic Explanation Techniques on Highly Imbalanced, Multiclass-Multioutput Classification Problem
Authors:

Syed Ihtesham Hussain Shah, Annette Ten Teije and José Volders
Abstract: Explainable AI (XAI) assist clinicians and researcher in understanding the rationale behind the predictions made by data-driven models which helps them to make informed decisions and trust the model’s outputs. Providing accurate explanations for breast cancer treatment predictions in the context of highly imbalanced, multiclass-multioutput classification problem is extremely challenging. The aim of this study is to perform a comprehensive and detailed analysis of the explanations generated by post-hoc explanatory methods: Local Interpretable Model-agnostic Explanation (LIME) and SHaply Additive exPlanations (SHAP) for breast cancer treatment prediction using highly imbalanced oncologycal dataset. We introduced evaluation matrices including consistency, fidelity, alignment with established clinical guidelines and qualitative analysis to evaluate the effectiveness and faithfulness of these methods. By examining the strengths and limitations of LIME and SHAP, we aim to determine their suitability for supporting clinical decision making in multifaceted treatments and complex scenarios. Our findings provide important insights into the use of these explanation methods, highlighting the importance of transparent and robust predictive models. This experiment showed that SHAP perform better than LIME in term of fidelity and by providing more stable explanation that are better aligned with medical guidelines. This work provides guidance to practitioners and model developers in selecting the most suitable explanation technique to promote trust and enhance understanding in predictive healthcare models.
Download
Paper Nr: 125
Title:

Analysis of Health Indicators for Heart Disease Based on Formal Concept Analysis
Authors:

Laura Xavier, Julio Neves, Luiz Zarate and Mark Song
Abstract: This study addresses the global concern of cardiovascular health by analyzing key risk factors such as high blood pressure, cholesterol levels, and smoking habits, which contribute to the onset of heart disease. Using Formal Concept Analysis (FCA), a mathematical framework for uncovering relationships in complex datasets, this research examines a health dataset of over 200,000 records to identify critical behavioral and health indicators related to cardiovascular problems. Although 80 association rules were extracted, 12 were selected for detailed analysis due to their significance in both risk and protective factors. Key findings reveal strong correlations between physical inactivity, poor dietary habits, and the likelihood of heart disease, providing actionable insights for healthcare professionals and policymakers. This study aims to deepen the understanding of cardiovascular risk factors and support the development of more effective prevention measures to improve global health outcomes.
Download
Paper Nr: 126
Title:

Using LLMs to Extract Adverse Drug Reaction (ADR) from Short Text
Authors:

Monika Gope and John Wang
Abstract: Adverse drug reactions (ADRs) are unexpected negative effects of a medication despite being used at its normal dose. Awareness of ADRs can help pharmaceutical companies refine drug formulations or adjust dosing guidelines to make medications safer and more effective. Twitter (X) can be a handy platform to extract unbiased ADR data from a large and diverse group of people. However, extracting ADRs from short texts such as tweets presents challenges due to the informal, noisy, and diverse nature of the text, which includes variations in user language, abbreviations, and misspellings. These factors make it difficult to accurately identify ADRs. Hence, it is important to identify the most effective strategies for extracting reliable ADR information. In this paper, we comprehensively evaluate various large language models (LLMs) and ML approaches for ADR extraction and detection. Using multiple ADR datasets and a range of prompt formulations, we compare the performance of each model. By systematically testing the effectiveness of these techniques across different combinations of models, datasets, and prompts, we aim to identify the most effective strategies for extracting reliable ADR information. Our study shows that LLMs excel in extracting ADRs, for example, with GPT-4 achieving an F1 score of 0.82, surpassing the previous ML methods of 0.64 for the SMM4H dataset. This indicates that LLMs are more effective and simpler alternatives to machine learning models for ADR extraction.
Download
Paper Nr: 131
Title:

Mobility: Promoting Health and Physical Activity in School Environments
Authors:

Mariana L. Moraes, Maria Luiza C. Santos, Maria das Graças P. Silva, Edyellen L. S. Oliveira, Rossana M. C. Andrade and Pedro Almir M. Oliveira
Abstract: Technology, although it has brought many advances in various social spheres, has contributed to the popularization of a sedentary lifestyle. The evolution of machines and the automation of daily processes have reduced the need for daily physical activities, contributing to a serious public health problem. Given this scenario, it is necessary to develop strategies that encourage adopting healthy practices in daily life. In this sense, this article investigates the implementation of the ”Mobility” application in high schools, aiming to promote physical activity among adolescents. By integrating technology with health promotion, Mobility stands out as a creative tool to reduce cases of sedentary lifestyles and improve the quality of life of young people.
Download
Paper Nr: 133
Title:

Deep Learning for ECG-Derived Respiration Using the Fantasia Dataset
Authors:

Lana Dominković, Biljana Mileva Boshkoska and Aleksandra Rashkovska
Abstract: In this paper, we explore a deep learning approach for extracting respiratory signals from electrocardiogram (ECG) data using the Fantasia dataset. We implemented a fully convolutional neural network model, inspired by the U-Net architecture, and designed to estimate respiratory signals from ECG data. The model incorporates convolutional layers, ReLU activations, batch normalization, max pooling, and up-sampling layers. Our deep learning model achieved an average correlation coefficient (CC) of 0.51 and Mean Squared Error (MSE) of 0.046, outperforming four out of six baseline signal processing algorithms based on the CC metric, and outperforming all signal processing algorithms based on the MSE metric. These findings demonstrate the effectiveness of deep learning in improving the accuracy and robustness of ECG-derived respiration (EDR). The research highlights the potential of advanced machine learning models for non-invasive respiratory monitoring and paves the way for future studies focused on exploring more complex architectures and broader datasets to further enhance performance and generalizability.
Download
Paper Nr: 136
Title:

Patient Trajectory Prediction: Integrating Clinical Notes with Transformers
Authors:

Sifal Klioui, Sana Sellami and Youssef Trardi
Abstract: Patient trajectory prediction from electronic health records (EHRs) is challenging due to the non-stationarity of medical data, the granularity of diagnostic codes, and the complexities of integrating multimodal information. While structured data, like diagnostic codes, capture key patient details, unstructured data, such as clinical notes, often hold complementary information overlooked by current approaches. We propose a transformer-based approach that integrates clinical note embeddings with structured EHR data for patient trajectory prediction. By combining these modalities, our model captures richer patient representations, improving predictive accuracy. Experiments on MIMIC-IV datasets show our approach significantly outperforms traditional models relying solely on structured data.
Download
Paper Nr: 144
Title:

Selection of Representative Instances Using Ant Colony Optimization: A Case Study in a Database of Newborns with Congenital Zika in Brazil
Authors:

Ana C. M. Gonçalves, Ludmila B. S. Nascimento, Ana L. P. Leite, Maria E. O. Brito, Erika G. de Assis, Henrique C. Freitas and Cristiane N. Nobre
Abstract: This article investigates congenital syndrome associated with the Zika virus (ZIKV) in newborns in Brazil, utilizing preprocessing techniques and machine learning to enhance its detection. The study proposes the Ant Colony Optimization (ACO) algorithm for instance selection in a database on ZIKV infections from 2016, during a period when Brazil faced a Zika outbreak linked to neurological complications such as microcephaly. The research compares the performance of ACO with five classification algorithms, demonstrating that ACO improved all evaluation metrics. The highest case concentration was observed in Brazil’s Northeast and Southeast regions. Although cases have decreased in 2024, it is essential to maintain monitoring and preventive actions. In summary, the results confirm the effectiveness of ACO in enhancing machine learning models and highlight the importance of clinical attributes in the early detection of congenital syndromes, recommending the use of updated databases for a better understanding of the impact of ZIKV, particularly in newborns.
Download
Paper Nr: 145
Title:

Smartphone Inertial Sensors in Gait Analysis: A Comparison with a Commercial Device
Authors:

Marco Oliveira, William Fröhlich, Rafael Baptista, Sandro Rigo and César Marcon
Abstract: Human gait analysis is a crucial tool in healthcare, providing valuable insights into an individual’s well-being, as various disorders and diseases can be detected through changes in walking patterns. This study aims to validate the gait sensing results obtained from a smartphone, an easily accessible and portable device, by comparing them with equivalent data from the G-Walk, a widely used commercial equipment. The goal is to assess the applicability and accuracy of the solution with the support of healthcare professionals, ensuring its effectiveness in clinical settings.
Download
Paper Nr: 147
Title:

Coordinates Transformed Signal Compression Method (CoTSiC): A Novel Algorithm for Tele-Medicine Applications
Authors:

Soham Pawar and Madhav Rao
Abstract: Robust Telemedicine refers to the provision of reliable remote medical services, which primarily depends on seamless transmission of either recorded signals or video information of patients in compressed form. A wide range of physiological signals which are typically seen in the display monitor of medical instruments including ECG, Blood Pressure, Oxygen levels, EEG signals and others are beneficial if remotely transmitted through reliable channels. Conventionally, the compression techniques applied to the signals are complex and compute-intensive, making it rarely viable at the remote patients’ end, where the compute infrastructure is scarcely available. To address this challenge, the paper introduces a lightweight compression algorithm specifically designed for these tele-healthcare applications. This work transforms the picture of the signal at the source into a compressed array of data points. This array is sent to the remote healthcare facility and then re-constructed into a minimalistic form of the signal. The proposed method offers a compression factor in the range of 3.87× to 2.82× for a variety of signals including EEG, ECG, and SPO2 signals. Additionally, an acceptable SSIM of above 92.10%, and PSNR of above 40 dB is characterized for the reconstructed image of different physiological signals investigated.
Download
Paper Nr: 152
Title:

Healful Dataset: Integrating Wearable Data with Self-Reported Quality of Life Assessments
Authors:

Pedro Almir M. Oliveira, Rossana M. C. Andrade, Pedro A. Santos Neto, Evilasio Costa Junior, Ismayle S. Santos, Victoria T. Oliveira, Wilson Castro and Leonan Carneiro
Abstract: This paper proposes a novel dataset – called Healful Dataset – correlating real data acquired from wearable health-tracking devices with Self-reported Quality of Life (SRQoL) measures collected using the WHOQOL-BREF questionnaire. Recently, increasing interest has been shown in using technology for Quality of Life (QoL) monitoring and improvement, significantly leveraging the Internet of Health Things (IoHT). Although several tools have been developed to quantify QoL, such as the SF-36 and WHOQOL-BREF, most are based on static and bothersome questionnaires rather than ubiquitous real-time data collection. Our database addresses this gap by integrating sensor-generated data with QoL assessment, enhancing the research path focused on intelligent models for QoL monitoring that use Machine Learning techniques to predict and improve QoL. In this paper, we describe the methodology used to build this database, the scenarios in which it can be applied, and discuss its relevance for future IoHT-driven health solutions toward improving people’s QoL through personalized monitoring and interventions.
Download
Paper Nr: 163
Title:

Understanding Stroke Risk Profiles in Middle-Aged Adults: A Genetic Algorithm-Based Feature Selection Aproach
Authors:

Ligia Ferreira de Carvalho Gonçalves, Caio Davi Rabelo Fiorini, Daniel Rocha Franca, Marta Dias Moreira Noronha, Mark Alan Junho Song and Luis Enrique Zárate Galvez
Abstract: Data mining and machine learning techniques have been widely used in the knowledge extraction process of medical databases, one highlight being their use to improve diagnostic systems. Decision trees are supervised black box machine learning models that, although simple, are easy to interpret. In this work, we propose the use of these techniques to describe the profile of middle-aged adults (40-59) diagnosed with stroke, a disease that in Brazil was one of the main causes of death in previous years. The genetic algorithm was applied to extract the best characteristics so that the Decision Tree algorithm could then be used in the database provided by the 2019 National Health Survey to obtain the most comprehensive rules and identify the most relevant attributes for describing the profile of these individuals. The conclusions indicate that the rules generated for middle-aged adults are mainly about routine habits, such as work or salt consumption.
Download
Paper Nr: 168
Title:

Using Machine Learning to Analyze the Impact of Lifestyle and Socioeconomic Factors on the Incidence of Depression Among Young Brazilians
Authors:

Thayris G. F. Rodrigues, Ariane C. B. da Silva and Cristiane N. Nobre
Abstract: Depression is a growing mental health problem among young people in Brazil, with factors such as socioeconomic and lifestyle conditions influencing its prevalence. This study investigates how variables such as education, family situation, and access to services impact the incidence of depression, using data from the National Health Survey (PNS) of the Brazilian Institute of Geography and Statistics (IBGE). Using machine learning algorithms such as Random Forest, XGBoost, SVM, and MLP, the analysis identified patterns among the factors, highlighting sleep problems and depressive feelings as the main determinants, with Recall above 70%. These results support the creation of more inclusive mental health policies.
Download
Paper Nr: 170
Title:

Application of Formal Concept Analysis for Identifying Depression Patterns in Adults in Brazilian National Health Survey
Authors:

Diogo Rodrigo dos Reis, Mark Alan Song and Cristiane Neri Nobre
Abstract: This study investigates the relationship between health status and lifestyle behaviors in relation to depression, utilizing data from the 2019 Brazilian National Health Survey (NHS). By applying Formal Concept Analysis (FCA) through dyadic analysis, we identify associations among health perceptions, health habits, social support, experiences of violence, and diagnosed conditions to explore depression patterns among Brazilian adults, including young adults, middle-aged adults, and older adults. The analysis reveals that a perception of good health is frequently linked to depression from the studied dataset. Gender-specific trends are also apparent, as females are more frequently diagnosed with depression compared to males, indicating a potential gender bias. Sleep problems and medicine to sleep were also associated with depression, as well as self-deprecating thoughts and some traces of violence experiences. Our findings emphasize the intricate interplay between health perceptions, health behaviors, and gender in understanding depression, highlighting the necessity for nuanced approaches in mental health assessments and interventions.
Download
Paper Nr: 178
Title:

Machine Learning-Based Clinical Decision Support Systems in Dementia Care
Authors:

Ritwik Raj Saxena and Arshia Khan
Abstract: Clinical Decision Support Systems (CDSSs) enabled by machine learning (ML), particularly those based on deep learning (DL), are revolutionizing dementia care by offering advanced capabilities that go beyond the capacities of manual CDSSs, rule-based CDSSs, and statistical CDSSs. This paper explores unique applications of ML in dementia care. It focuses on areas where ML-based, especially DL-based (neural network-based) CDSSs currently excel and can potentially be of relevance. Unlike conventional CDSSs, which evidently struggle with the complexities of large, heterogeneous datasets, ML models, particularly DL-based ones, are capable of better identifying hidden patterns and subtle relationships across diverse genetic and multi-omic, clinical, behavioural, socioeconomical and cultural, and environmental data. These systems also extend their utility beyond clinical decision-making and caregiver wellbeing through tailored support recommendations and aiding hospital administrators in resource mobilization, staff augmentation, and policy formulation. However, challenges such as model interpretability, extensive data requirements, and infrastructure limitations must be addressed. This article highlights the importance of a collaborative approach, where various stakeholders in dementia come together to pool data and recommendations that would assist in inculcating comprehensiveness and inclusivity in future CDSSs. We hypothesize that as DL continues to showcase its decided prowess in the arena of decision-making, its applications in CDSSs will keep playing an exceedingly pivotal role in advancing the efficacy of dementia care, improving patient outcomes, and shaping the future of healthcare.
Download
Paper Nr: 203
Title:

Assessing Goal Disengagement Using a Digital, Card-Based Game: A Proof of Concept Study
Authors:

Sebastian Unger, Hana Minařík and Thomas Ostermann
Abstract: The distinction between goal engagement (GE) and goal disengagement (GD) as central psychological processes is supported by several theories of developmental regulation. However, although there has been research on both, research on GD has been rather neglected, especially when it comes to behavioral methods for its assessment. The objective of this paper, therefore, is to evaluate the feasibility of such a behavioral method by placing a homogeneous group of participants in a situation where they need to distinguish whether the effort to solve a digital, card-based game leads to successful goal achievement or to frustration. The data from this group revealed no significant differences in the participants' behavior over the course of the game. Nonetheless, some tendencies in the number of repetitions and the number of cards collected until the occurrence of a GD could be found when differentiating between participants who adhered to their goals more persistently and those who disengaged more frequently. Overall, the game may have potential for both replacing previous assessment methods and identifying suitable individuals for long-term rehabilitation and behavioral therapies, but further research is required for application in a clinical setting.
Download
Paper Nr: 204
Title:

Czech Salivary Gland Database
Authors:

Vojtěch Jelínek, Petr Brůha, David Kalfeřt and Pavel Nový
Abstract: Salivary gland tumors require comprehensive data collection and analysis to support clinical decision-making, yet existing databases need more focus on specific tumor-related data and visualization tools. This absence hinders oncologists’ ability to track patient outcomes effectively and identify potential prognostic indicators. To address this, we developed the Czech Salivary Gland Database (CSGDB), a specialized clinical application designed to manage patient data and provide visual analytics. The database includes secure and anonymized data handling alongside Kaplan-Meier survival analysis for outcome visualization. Deployed at the University Hospital in Motol, CSGDB empowers healthcare professionals with enhanced tools for tracking and analyzing patient progress, ultimately contributing valuable data and insights to the field of head and neck oncology.
Download
Paper Nr: 206
Title:

Relationships Between Central Quality Assurance Criteria for the Assessment of Statutory Health Insurance Patients for Further Development Using Medical Informatics Methods
Authors:

Mareike Burmester, Paul-Ulrich Menz, Vera Ries, Klaus-Peter Thiele, Bernhard van Treeck and Reinhard Schuster
Abstract: In recent years, a comprehensive system has been developed both in terms of content and IT for quality assurance in the assessment for the granting of benefits for persons with statutory health insurance by the Medical Advisory Boards, which has also been enshrined in German legislation. In addition to not insignificant formal criteria and criteria relating to specific assessment areas, four criteria relevant to the entire assessment spectrum are evaluated in detail. One- and two-dimensional criteria provide an overview as an introduction to the topic. Similar to the procedure in image processing, linear optimization methods are used to infer relevant intervals of the detailed parameters from row and column totals. Using correlations and partial correlations, the relationship between the central quality criteria is shown. Methods of spherical trigonometry are generalized. For each of the three sides of the quadrilateral of the four central criteria, it is of central importance that the partial correlations are greater or smaller than the correlations overall. This is determined by the modulus value, which in the application under consideration produces the same results on all sides of the tetrahedron under consideration.
Download
Paper Nr: 212
Title:

Multiscale Entropy Analysis of Continuous Glucose Monitoring Data: A Comparative Study of Diabetic and Healthy Populations
Authors:

Cleber França Carvalho, Thilini Savindya Karunarathna and Zilu Liang
Abstract: The advent of continuous glucose monitoring (CGM) has made it possible to measure glucose frequently in daily life. This availability of glucose time series enables advanced analysis to uncover patterns in glycaemic dynamics that were previously undetectable with traditional blood-sample-based measurements. One such analytical method is multiscale entropy (MSE), which assesses the complexity of time series data across varying time scales. In this study, we performed a comparative analysis of MSE across three cohorts: individuals with type 1 diabetes (T1D), type 2 diabetes (T2D) and prediabetes (PRED). Our goal was to identify potential differences in glucose dynamics across these groups. We applied three base entropies, including approximate entropy (ApEn), attention entropy (AttnEn) and dispersion entropy (DispEn). We found that AttnEn and DispEn were useful in distinguishing between individuals with diabetes (both T1D and T2D) and those with prediabetes, whereas ApEn did not show significant discriminative power. Furthermore, we observed no substantial differences between T1D and T2D in terms of their MSE profiles. These results suggest that MSE, with appropriate base entropy measures, holds promise as a tool for developing biomarkers to differentiate between diabetes and prediabetes. Future studies could explore additional base entropy measures and analysing larger, more diverse datasets.
Download
Paper Nr: 214
Title:

Recognising Care-Related Activities with a Convolutional Neural Network Using Inertial Measurement Units
Authors:

Fenja T. Hesselmann, Jannik Fleßner, Alexander Pauls, Vanessa Cobus, Frauke Koppelin and Frank Wallhoff
Abstract: Sensor-based human activity recognition is a growing field of research. In addition to recognising everyday movements, situation-dependent activities can also be detected. This paper therefore aims to detect care-specific movements. For this purpose, 13 different nursing activities were recorded with Inertial Measurement Units (IMUs) worn on the body. In this paper, we present an approach on how the sensor data can be used for recognition. Convolutional neural networks were used for classification. The focus of this work is on two different fusion approaches of the data to check which approach achieves better results. In the first approach, all data is fused at the beginning, while in the second one, a separate pipeline is designed for each sensor and fused later. The results show that a later fusion technique provides a better F1 score of 90.2 % compared to a model that considers all signals from the beginning (F1 score: 82.5 %).
Download
Paper Nr: 223
Title:

Expertise versus Data: Comparison of Expertise Based Process Models to Process-Mining Models of Surgery Under General Anesthesia
Authors:

Hugo Boisaubert, Chloé Grivaud, Antoine Bouchet, Corinne Lejus-Bourdeau and Christine Sinoquet
Abstract: Digital tools accessible for healthcare are often based on models representing a medical process and learned from medical data. Unfortunately, those data are protected by privacy regulation and therefore are quite rare. This rarity leads to process models mainly based on the expertise of caregivers. Those expertise-based model and data-based models are rarely compared to show their common characteristics and differences. When both model can be produced for the same situation multiple questions arise. Should the expertise-based model be invalidated if it is not in full conformity to the data-based model ? Are those models' characteristics the same? In this article, we present a comparison of expertise-based models and data-based models produced for a surgery under general anesthesia with 204 real cases. We conducted a process mining algorithm performance comparison on our specific real data to identify the most promising learning method. Then we compared the produced data-based models to the expertise-based models with some metrics. The comparison results show strong differences between the two types of models, the expertise-based model is very much smaller than the data-based model, but we have noticed that the expert-based model is included in a data-based model. Therefore, the main difference between the two models appears to be on a level of abstraction.
Download
Paper Nr: 225
Title:

Leveraging LLMs and RAG for Schema Alignment: A Case Study in Healthcare
Authors:

Rishi Saripalle, Roopa Foulger and Satish Dooda
Abstract: In the quest to achieve digital health and enable data-driven healthcare, health organizations often rely on multiple third-party vendor solutions to monitor and collect patient health and related data, specifically outside organizations' control, such as home setting, which is later communicated to the organization’s information systems. However, the reliance on multiple vendor solutions often results in fragmented data structures, as each vendor solutions system follows its non-standard data model. This fragmentation complicates the data integration, creating barriers to seamless data exchange and interoperability, which is essential for data-driven healthcare. Recent advancements in Large Language Models (LLMs) have great potential to analyze data models and generate rich contextual-semantic metadata for the model, useful for identifying mappings between disparate data structures. This preliminary research explores the adoption of LLMs in combination with the Retrieval-Augmented Generation (RAG) approach to facilitate structural alignment between disparate data models. By semi-automating the schema alignment process—currently a labor-intensive task—LLMs can streamline the data integration of heterogeneous data models, enhancing efficiency by reducing the developer’s time and manual effort.
Download
Paper Nr: 232
Title:

D-Care: A Multi-Tone LLM-Based Chatbot Assistant for Diabetes Patients
Authors:

Awais Khan Nawabi, Janos Tolgyesi, Elena Bianchi, Chiara Toffanin and Piercarlo Dondi
Abstract: Diabetes is a common chronic illness projected to increase significantly in the coming years. Managing diabetes is complex, requiring patients to frequently adjust their treatments and lifestyles to prevent complications. Awareness and adherence to healthy habits are thus essential. Artificial Intelligence (AI) can assist in this effort. Recent advancements in Large Language Models (LLMs) have enabled the creation of effective chatbots to support patients. However, despite their growing use, there are still a few formal user studies on LLMs for diabetes patients. This study aims to investigate the ability of an LLM-based chatbot to provide useful and understandable information to potential patients. Specifically, the goal was to examine how variations in language and wording affect the comprehension and perceived usability of the chatbot. To this end, D-Care, a chatbot assistant based on OpenAI’s ChatGPT-4o, was developed. D-Care can generate answers in four different tones of voice, ranging from elementary to technical language. A user study with 40 participants showed that changes in tone can indeed impact the system’s comprehension and usability.
Download
Paper Nr: 239
Title:

Mobile Data Collection for Depression Analysis: An App Framework for Monitoring Mood and Depression Using Smartphone and Wearable Data
Authors:

Eliane Schröter, Franziska Klein, Patrick Elfert, Fynn Bredehorn, Julien Räker, Frerk Müller-von Aschwege and Andreas Hein
Abstract: Depression is a leading cause of disability worldwide, affecting around 5% of the global adult population. To address this problem, researchers are exploring methods for early detection of relapses, mood swings and their relation with health data and external influences. The aim of the present study was to evaluate the usability and feasibility of a mobile application designed for active and passive data collection, with potential future applications for improving mental healthcare through a virtual therapy assistant. The application allows users to self-report their mood, complete PHQ-9 questionnaires, and track measures such as sleep, physical activity, location, smartphone usage, and social media engagement. A six-week pilot study was conducted with 22 healthy participants (68% male, 32% female). Participants recorded their mood three times a day and completed weekly mental health assessments. Results showed that the application effectively collected relevant data and was user friendly. However, limitations included reliance on self-reported data, short study duration, and occasional technical issues with data collection. Despite these limitations, the study showed that it is possible to use smartphones and wearable technologies to monitor mental health, laying the foundation for future developments in digital therapeutic interventions and personalized healthcare through app-based virtual therapy assistants.
Download
Paper Nr: 247
Title:

Using Artificial Intelligence and Large Language Models to Reduce the Burden of Registry Participation
Authors:

James P. McGlothlin and Timothy Martens
Abstract: Health care disease registries and procedural registries serve a vital purpose in support of research and patient quality. However, it requires a significant level of clinician effort to collect and submit the data required by each registry. With the current shortage of qualified clinicians in the labor force, this burden is becoming even more costly for health systems. Furthermore, the quality of the abstracted data deteriorates as over-worked clinical staff review and abstract the data. The modern advancement in electronic medical records has actually increased this challenge by the exponential growth in data volume per patient record. In this study, we propose to use large language models to collect and formulate the registry data abstraction. For our initial work, we examine popular and complicated patient registries for cardiology and cardiothoracic surgery. Initial results demonstrate the promise of artificial intelligence and reenforce our position that this technology can be leveraged.
Download
Paper Nr: 250
Title:

GLOW-ENV: A Dual-Data IoE-Based Approach for Integrating Glucose and Environmental Data into a Diabetes Recommendation System
Authors:

Juan F. Gaitán-Guerrero, Carmen Martínez-Cruz, José L. López, Francisco Mata-Mata, Luis G. Pérez-Cordón, José-María Serrano, Juana M. Morcillo-Martínez, Ángeles Verdejo-Espinosa, Juan C. Cuevas-Martínez, Raquel Viciana-Abad, Pedro J. Reche-López, José M. Pérez-Lorenzo, David Díaz-Jiménez and Macarena Espinilla
Abstract: This paper introduces GLOW-ENV, an intelligent Internet of Everything (IoE)-driven mobile application designed with the objective of integrating real-time glucose monitoring data and environmental metrics to enhance diabetes care and management. The proposed IoE ecosystem integrates a continuous glucose monitoring with a personalized Artificial Intelligence model designed to predict glycemic fluctuations in a near-future. Additionally, GLOW-ENV integrates a rule-based recommendation system to dynamically adapt its suggestions based on contextual glucose and environmental data. This framework advances personalized diabetes care, contributing to their progression and well-being offering valuable insights and improving decision-making.
Download
Paper Nr: 253
Title:

A Study on mHealth Adherence for Bipolar Disorder: A Case Study with the BraPolar2 Application
Authors:

Abel González Mondéjar, Luiza Oliveira Régnier, Greis Francy M. Silva-Calpa and Daniel C. Mograbi
Abstract: Bipolar disorder (BD) is a mental illness that affects 40 million people worldwide. Fluctuations in mood, activity, and self-awareness mark this condition. Studies using mHealth applications to monitor people with BD have shown promising results in the early detection of these fluctuations; however, they usually require participants to complete daily tasks in the app, which causes them to abandon the study and compromises the quality of the research. This paper explores the adherence to BraPolar2 mHealth through a set of development strategies. To identify the aspects that lead patients with BD not to complete daily data in an mHealth application and the factors that motivate them to use the application as a habit, we conducted qualitative research with BraPolar2 mHealth. Nine people with BD participated in the study and used BraPolar2 for more than 3 months, answering a semi-structured interview. The results show that users can fill in all the data quickly and begin to pay more attention to their mental health daily. The paper contributes by demonstrating how a simplified interface in mHealths, coupled with qualitative research, can lead to the participation of mHealth applications for mental health follow-up, allowing an improved follow-up in next studies.
Download
Paper Nr: 255
Title:

Profiling Cancer Patients in Telemedicine: A Case Study in Chilean Private Healthcare
Authors:

Matías Cornejo, Esteban Chiu, Sebastián Valderrama, Sebastián Mondaca and Eric Rojas
Abstract: Cancer is a devastating disease that affects patients and places a growing burden on healthcare systems globally. In recent years, cancer diagnoses have increased, exacerbating the challenges faced by both patients and healthcare providers. With the advancement of innovative technologies and rising demands for care, telemedicine has become a viable option for ensuring continued treatment, including for cancer patients. However, little is known about the characteristics and behaviors of these patients. This study aims to characterize cancer patients accessing telemedicine services at a private healthcare institution in Chile from 2020 to 2023, providing deeper insights into their profiles. Preliminary findings indicate a decline in adult patients and telemedicine appointments between 2020 and 2023. Similarly, pediatric patients experienced a decrease in telemedicine use from 2020 to 2022, followed by a slight increase in 2023. Understanding the profiles and behaviors of cancer patients utilizing telemedicine is crucial for improving their healthcare journeys. By analyzing their experiences, healthcare providers can enhance the allocation and management of resources, ensuring more effective and personalized care. This characterization also supports the development of strategies to optimize telemedicine services, improving outcomes for cancer patients in a rapidly evolving healthcare landscape.
Download
Paper Nr: 258
Title:

U-Net in Medical Imaging: A Practical Pathway for AI Integration in Healthcare
Authors:

Martin Kryl, Pavel Košan, Petr Včelák and Jana Klečková
Abstract: As AI transforms medical imaging, this paper positions U-Net as a practical and enduring choice for segmentation tasks in constrained clinical environments. Despite rapid advancements in architectures like transformers and hybrid models, U-Net remains highly relevant due to its simplicity, efficiency, and interpretability, particularly in settings with limited computational resources and data availability. By exploring modifications such as residual connections and the Tversky loss function, we argue that incremental refinements to U-Net can bridge the gap between current clinical needs and the potential of more advanced AI tools. This paper advocates for a balanced approach, combining accessible enhancements with hybrid strategies, such as radiologist-informed labeling and advanced preprocessing, to ensure immediate impact while building a foundation for future innovation. U-Net’s adaptability positions it as both a cornerstone of today’s AI integration in healthcare and a stepping stone toward adopting next-generation models.
Download
Paper Nr: 261
Title:

Digital Medication Prescription System with JSON
Authors:

Liverson Paulo Furtado Severo and Jean Everson Martina
Abstract: The rapid evolution of digital health solutions, accelerated by the COVID-19 pandemic, highlighted the necessity for secure and efficient electronic prescription and medication dispensing systems. This paper presents a study on integrating HL7 FHIR standards and JAdES signatures to facilitate the digitalization of healthcare documentation. By addressing key challenges such as interoperability, data volatility, and security, the research proposes a framework that ensures the authenticity and integrity of electronic prescriptions—Emphasizing the importance of self-contained digital documents that eliminate reliance on external references to enhance the reliability of health information exchange in Brazil. Furthermore, it outlines the legal implications of electronic signatures in Brazil, advocating for compliance with national standards to ensure the legal validity of digital prescriptions. The findings indicate that the proposed solutions not only streamline healthcare processes but also foster a gradual transition from traditional paper-based systems to a robust digital infrastructure, ultimately improving patient care and operational efficiency in the healthcare sector.
Download
Paper Nr: 266
Title:

Overview of Best Communication Practices in Inclusive Spaces for People with Disabilities: A Qualitative Study in Portuguese-Speaking Countries
Authors:

Francisca Rocha Lourenço, Rita Oliveira and Oksana Tymoshchuk
Abstract: The inclusion of People with Disabilities (PwD) is still a challenge in Portuguese-Speaking Countries (PSC), requiring continuous efforts to ensure their full participation in society. The goal of this study was to identify and understand the best communication practices adopted and recognised by Inclusive Spaces (IS) in PSC, aiming to contribute to the promotion of inclusion of PwD. Using a qualitative approach, 16 semi-structured interviews were conducted with representatives of various IS in Portugal, Brazil and Angola. The results highlighted a set of communication practices applied in different contexts, including inclusive communication, use of social media and sharing of real-life testimonies, as practices that strengthen public engagement and the dissemination of the services offered by IS. Strategies such as the creation of inclusive and accessible content and the use of channels, such as websites and face-to-face events, were highlighted for their ability to increase the visibility of spaces, reach wider audiences, and reinforce their role in society. The findings emphasise the importance of effective communication in strengthening IS and ensuring that their services reach PwD. By outlining a set of good practices, this study provides initial guidelines for improving communication in IS, contributing to the promotion of inclusion of PwD.
Download
Paper Nr: 271
Title:

Comparison of Bluetooth Low Energy (BLE), Wi-Fi, Serial and 5G in IoMT
Authors:

Nina Pearl Doe, Stefan Scharoba, Marc Reichenbach and Christian Herglotz
Abstract: With the inception of Industry 4.0, incorporating technologies like the Internet of Things (IoT) into healthcare has become essential. This integration is commonly referred to as the Internet of Medical Things (IoMT). The IoMT is the connection of medical devices using wired or wireless data transmission technology to allow data exchange with the goal of improving the overall healthcare delivery. Despite the numerous advantages that IoMT brings into the healthcare process, there are potential performance challenges that may occur if factors such as data quality and reliability of the IoT devices in different environmental settings are not properly considered. The purpose of this paper is to analyse the performance of connected medical IoT devices that are used for heartrate monitoring based on the aforementioned factors. The setup of the IoMT consists of sensor nodes, which transmit the Electrocardiogram (ECG) data through a multi-protocol gateway to a central server for further data processing. This paper presents the performance analysis of the comparison of four communication technologies: Serial (UART), Bluetooth Low Energy (BLE), Wi-Fi, and 5G NR for real-time ECG monitoring applications, while taking notice of environmental factors that may affect performance. The sensor data transmission is evaluated based on round trip time (RTT) latency, ensuring a desirable throughput and minimal or no data loss. The data readings were taken at varying distances (0.1m to 17m) and sampling rates (300Hz and 1000Hz). The experimental results show that while Serial communication achieves the lowest latency (3.96ms - 4.37ms), Wi-Fi demonstrates consistent Gateway-Server performance (40ms - 60ms RTT), 5G excels in short-range communication (1.8ms - 2.0ms Sensor Node-Gateway RTT), and BLE provides balanced performance (4.86ms - 7.57ms latency). Wi-Fi performed better in long-range scenarios (43.48ms -66.23ms RTT) and maintaining stable performance at longer ranges while 5G shows superior performance in short-range, high-frequency scenarios.
Download
Paper Nr: 276
Title:

Identifying an Autoinflammatory Syndrome Cohort Using Natural Language Processing with Electronic Medical Record Data
Authors:

Maranda Russell, Aleksander Lenert, Katherine Liao, Tianrun Cai and Sujin Kim
Abstract: Autoinflammatory syndromes (AIS) are rare inflammatory disorders with diverse and severe manifestations, making their clinical outcomes and phenotypes poorly understood. This study developed and validated machine learning algorithms incorporating clinical natural language processing (cNLP) and electronic medical record (EMR) data to identify AIS cases. Patients were filtered using relevant billing codes, medications, and ICD-9/-10 codes for conditions such as adult-onset Still’s disease, Behcet's disease, and familial Mediterranean fever. Machine learning models—adaptive lasso penalized logistic regression (ALASSO), support vector machine (SVM), and random forest (RF)—utilized structured codes and cNLP-extracted features. Of 206 patients screened, 61 (29.6%) were confirmed AIS cases after manual review. SVM (AUC=0.954) and RF (AUC=0.948) outperformed ALASSO (AUC=0.94). A total of 44 features, including ICD codes for arthritis and Behcet's disease and cNLP-derived concepts such as periodic fever, oral lesions, and colchicine treatment, were predictive of AIS. This study demonstrates the feasibility of combining structured and unstructured EMR data for AIS identification, providing a scalable framework for phenotyping rare diseases and advancing outcomes research.
Download
Paper Nr: 300
Title:

Comparison of Different User Interfaces for 360-Degree Videos in VR-Based Healthcare Education
Authors:

Yan Hu, Jessica Berner, Veronica Sundstedt and Ivan Perlesi
Abstract: Extended reality (XR) technology has been increasingly used in many areas, one being healthcare. This paper presents a pilot study comparing two 360-degree virtual reality (VR) healthcare applications. The applications were evaluated by eight nursing students who evaluated both interfaces based on the User Experience Questionnaire (UEQ), the System Usability Scale (SUS), and the Simulator Sickness Questionnaire (SSQ). Results show that both applications could provide a positive user experience and high usability, with some improvements shown in the second version of the application. The SSQ scores also showed that minimal motion sickness occurred. Overall, all participants thought the VR-based education provided an innovative alternative to traditional education scenarios.
Download
Paper Nr: 309
Title:

Assessing Dietary Protein Intake: A Comparative Study of Two Consumer Mail-In Nutrition Test Kits
Authors:

Zilu Liang
Abstract: This study aimed to compare two consumer-grade mail-in nutrition test kits, Flemi Check and VitaNote, in measuring protein intake and identifying protein deficiencies. A total of 18 subjects (10 male, 8 female) aged 19 to 36 years participated. Descriptive statistics revealed that most subjects consumed between 60 and 80 grams of protein per day, slightly below the recommended 80 grams. The Flemi Check test identified 15 subjects as protein-deficient, while the VitaNote test identified 11. A significant disparity in protein consumption measurements was found, with the Flemi Check consistently underestimating protein consumption compared to the VitaNote test for 16 out of the 18 subjects, with a mean difference of 17.11 grams. However, both kits showed good agreement in estimating the recommended daily protein intake, with only a 2-gram difference. Given the high precision of the VitaNote test, the Flemi Check may not be considered as a reliable tool for assessing protein intake.
Download
Paper Nr: 313
Title:

Multidimensional Correlations in the Implementation in Medical Informatics and Their Statistical and Epidemiological Evaluations in the Quality Assurance in the Medical Advisory Board in Germany
Authors:

Vera Ries, Reinhard Schuster, Paul-Ulrich Menz, Klaus-Peter Thiele, Bernhard van Treeck and Mareike Burmester
Abstract: In quality assurance within the Medical Advisory Board in Germany, the structures that are primarily organised by federal state are are being networked nationwide. The aim is to implement a sufficiently standardised nationwide assessment. The differing regional starting points are simply due to the different mandates from the health insurance funds. In up to four levels of supra-regional interaction, a standardised assessment is being steadily improved in the implemented process. This process is being improved on a continuous basis. Statistical and epidemiological evaluations with proven health economic measures and graph-theoretical methods using the Mathematica software system from Wolfram Research.
Download
Paper Nr: 319
Title:

Design and Implementation of an Open Clinical Trials Platform Using HL7® FHIR® Within the Orthokids-Project
Authors:

Anne Grohnert, Michael John, Benny Häusler, Christian Giertz, Ben Kraufmann, Stefan Klose and Conrad Klaus
Abstract: Design and implementation of real care processes is a complex task, where the information model is central and serves as a basis for further implementation of the overall IT system. In this contribution, we describe the process of modeling study-specific structures on the base of FHIR in the OrthoKids project. The OrthoKids project is a clinical study aimed at establishing an orthopedic preventive medical examination for children. We explain how FHIR was used for modeling the data and in what form it is used, ending in summarizing our experiences we made when modeling the OrthoKids study with FHIR.
Download
Paper Nr: 321
Title:

Preparing Ultrasound Imaging Data for Artificial Intelligence Tasks: Anonymisation, Cropping, and Tagging
Authors:

Dimitrios Pechlivanis, Stylianos Didaskalou, Eleni Kaldoudi and George Drosatos
Abstract: Ultrasound imaging is a widely used diagnostic method in various clinical contexts, requiring efficient and accurate data preparation workflows for artificial intelligence (AI) tasks. Preparing ultrasound data presents challenges such as ensuring data privacy, extracting diagnostically relevant regions, and associating contextual metadata. This paper introduces a standalone application designed to streamline the preparation of ultrasound DICOM files for AI applications across different medical use cases. The application facilitates three key processes: (1) anonymisation, ensuring compliance with privacy standards by removing sensitive metadata; (2) cropping, isolating relevant regions in images or video frames to enhance the utility for AI analysis; and (3) tagging, enriching files with additional metadata such as anatomical position and imaging purpose. Built with an intuitive interface and robust backend, the application optimises DICOM file processing for efficient integration into AI workflows. The effectiveness of the tool is evaluated using a dataset of Deep Vein Thrombosis (DVT) ultrasound images, demonstrating significant improvements in data preparation efficiency. This work establishes a generalizable framework for ultrasound imaging data preparation while offering specific insights into DVT-focused AI workflows. Future work will focus on further automation and expanding support to additional imaging modalities as well as evaluating the tool in a clinical setting.
Download
Paper Nr: 322
Title:

LOINC Mapping Experiences in Italy: The Case of Friuli-Venezia Giulia Region
Authors:

Maria Teresa Chiaravalloti, Grazia Serratore, Fabio Del Ben and Agostino Steffan
Abstract: Interoperability in healthcare requires accurate data exchange and interpretation across systems, making standard terminologies essential for achieving semantic interoperability. This paper presents the approach adopted by the Friuli Venezia Giulia Region in Italy to implement LOINC, the most widely used standardized coding system for laboratory tests, into the electronic Laboratory Reports of five hospitals. Mapping was conducted manually by physicians using RELMA, supported by training and guidance from LOINC Italy experts. The validation process involved a dual-review procedure to ensure semantic accuracy but also to face issues, such as implicit or incorrect information in local catalogues and the complexity of some specialties. Collaboration among clinical staff, LOINC experts, and IT professionals proved essential in overcoming these issues. As a result, over 7,000 local tests were mapped to LOINC, and 675 new codes for unrepresented concepts were requested, thus creating a regional LOINC knowledge base. This experience highlights the importance of training, support, and integrated management in adopting LOINC, as these elements are crucial for a standardization process that enhances data traceability, minimizes errors, and supports semantic interoperability. Additionally, this experience could be an example for other healthcare systems aiming to standardize laboratory tests and achieve meaningful data exchange.
Download
Paper Nr: 331
Title:

Eyes as Windows to the Heart: Predicting Heart Rate from Pupillometric Features
Authors:

Kevin Kristofer Kosasih, Carl Daniel Karlsson, Thilini Savindya Karunarathna and Zilu Liang
Abstract: Heart rate is a key indicator of health, typically measured through skin-contact methods such as electrocardiograms (ECG) or photoplethysmograms (PPG). However, these methods may not be comfortable for everyone, prompting interest in non-contact alternatives. Eye tracking presents a promising solution, as the autonomic nervous system links the eyes to heart rate. This research develops heart rate prediction models based on pupillometric features. We conducted data collection experiments to build a dataset of multi-modal measurements of pupillometric data and heart rate from 10 subjects at high sampling rates. Several regression models, including linear regression, ridge regression, random forest regression, and XGBoost regression, were trained on the dataset. The random forest model achieved the best performance with a R2 of 0.457 and a root mean square error (RMSE) of 9 beats per minute, representing a 52.3% improvement over the state-of-the-art. Future work should focus on expanding the dataset, refining feature extraction and selection, and incorporating 3D pupillometric data to enhance model accuracy and applicability.
Download
Paper Nr: 333
Title:

Reconsidering AHI as an Indicator of Sleep Apnea Severity: Insights from Mining Large, Longitudinal Sleep Datasets
Authors:

Nhung H. Hoang and Zilu Liang
Abstract: Sleep apnea remains a key area of sleep research, with the Apnea-Hypopnea Index (AHI) widely used to assess its severity. This study evaluated whether AHI is truly the best indicator of sleep apnea and identified its limitations. Using the Sleep Heart Health Study and Wisconsin Sleep Cohort datasets, which provide large, longitudinal data, we also explored survey data on demographics, physiology, and daily behaviors—often overlooked in polysomnography-based studies. The results indicate that AHI may be a good indicator for mild or moderate sleep apnea, but not necessarily for normal or severe cases. We highlight some trends that can be seen from longitudinal data. Additionally, using contrast set mining method, we identified key risk factors for cardiovascular disease, including age, snoring, and smoking behavior. These results underscore the importance of considering AHI’s limitations and incorporating additional factors for more accurate sleep apnea diagnosis and risk assessment.
Download
Paper Nr: 352
Title:

Physiology-Guided Blood Glucose Predictive Model Using Minimal Blood Glucose Dynamics
Authors:

Sarala Ghimire, Turgay Celik, Martin Gerdes and Christian W. Omlin
Abstract: Modelling blood glucose and insulin dynamics using mathematical equations requires a deep understanding of individual physiology and relying on numerous predefined parameters necessitating extensive clinical and personal data, making direct use of these models for blood glucose prediction computationally intensive and inaccurate. Though data-driven models are more efficient and require no individual physiology, they produce predictions that are inconsistent with known glucose-insulin interactions. Thus, this study aims to investigate the potentiality of physiological models integrated with data-driven approach for predicting blood glucose level. It intends to extract simple physiological dynamics of blood glucose kinetics and incorporate them into a data-driven model, with less reliance on detailed individual data. The result demonstrated that the model integrating physiological modelling of insulin and meal absorption significantly improved the performance particularly in larger window size that enabled the model to better capture longer-term trends and temporal dependencies.
Download
Paper Nr: 39
Title:

Innovation in Geriatric Care: An AI Assistant with LLM Integration Based on Health Guidelines
Authors:

Juliana Basulo-Ribeiro, Nádia C. G. Matos, Sabrina Magalhães Araujo, Nuno Capela, Francisco Bischoff, Leonor Teixeira and Ricardo Cruz-Correia
Abstract: With the advance of artificial intelligence and natural language processing technology, a new tool is standing out in the field of understanding and generating natural language in a sophisticated way: the Large Language Model (LLM). According to several authors, LLMs can be used for various types of medical cases, providing access to different sources of information, and have opened up countless opportunities in the healthcare sector. This work aims to share the lessons learned during the process of developing an LLM-based assistant aimed at specific pathologies that are more prevalent in the elderly in order to support caregivers, whether they are private individuals, home care organizations, nursing homes or others. This study has a significant potential impact on the community by providing access to detailed information on developing an LLM-based assistant.
Download
Paper Nr: 51
Title:

Impact of Balancing and Regularization on the Semantic Segmentation of Gleason Patterns
Authors:

Eduardo Henrique S. Paraíso and Alexei M. C. Machado
Abstract: This study investigates the impact of class balancing and regularization on improving the diagnostic agreement in prostate histological images. The U-Net models applied to the Prostate Cancer Grade Assessment dataset reveal that class balancing combined with traditional loss functions contributes to an increase of up to 6 percentage points in image agreement. Combining balancing and Focal Loss can increase image classification agreement by an average of 13 percentage points compared to using an imbalanced dataset with traditional loss functions. Notably, distinguishing between Gleason patterns 3 and 4 in medical image analysis is crucial, as this distinction not only directly influences clinical decisions and the prognosis of prostate cancer patients but also emphasizes the need for careful interpretation of the data.
Download
Paper Nr: 55
Title:

A Flexible and Open-Source Tool for Genetic Variant Annotation
Authors:

Andrea Bombarda, Matteo Bellini, Maria Iascone and Domenico Fabio Savo
Abstract: Advances in genomic research have significantly enhanced our understanding of the genetic factors influencing human health. A key output of this research are VCF (Variant Call Format) files, which document genetic variations detected through DNA sequencing. These files, however, provide limited information, making it challenging to interpret the biological significance of the variants without additional data. Annotation, the process of enriching VCF files with information from publicly available biomedical datasets, is essential for facilitating variant interpretation in research. In this paper, we present VCFAnnotator, a tool developed to adapt ANNOVAR software used in genetic research, enabling the annotation of entire directories with a single command and facilitating the use of any relevant external database. Additionally, VCFAnnotator offers the ability to scrape the various websites of the biomedical databases in use, ensuring that the researchers remain informed of any updates.
Download
Paper Nr: 75
Title:

The Role of Online Communities in Promoting Physical Activity: A Survey on User Preferences and Perceived Impact
Authors:

Jennifer Hachiya
Abstract: The primary objective of this online survey is to understand differences in user profile, user preferences and perceived impact among the European population. The sample groups were based on the most recent report of the European country with the highest and lowest levels of physical activity (PA). The cross-sectional online survey population of Portugal residents and Finland residents was selected by simple random sampling. Responses were collected from the open-source tool LimeSurvey. IBM Statistical Package for Social Sciences Statistics was used to analyse the acquired data. A total of 538 responses were considered with 48.4% of respondents residing in Portugal, and 51.4% residing in Finland. About 38.5% of the general survey population regularly practice exercise, and 39.7% regularly engage in PA. Regarding the level of online community experience, responses were distributed between medium, moderately low, and very low. Overall, there is a significant relationship between both sample groups when it comes to PA, common emotions using online communities, user perception, preferences and openness. Our survey results provide evidence to support that country of residence is related to user PA and highlight the importance of considering demographic factors to understand general population lifestyle choices.
Download
Paper Nr: 111
Title:

A Multiple Source Data Collection and Integration Paradigm for the Creation of a Dynamic COPD Data Mart
Authors:

Giulio Pagliari, Agni Delvinioti, Nicoletta Di Giorgi, Maria Vittoria De Girolamo, Angela Nervoso, Francesco Macagno, Carlotta Masciocchi, Stefano Patarnello and Alice Luraschi
Abstract: The creation of dynamic data marts in a hospital environment is challenging due to the number of different data sources, the heterogeneity of data formats and the availability of structured datasets. Other than identifying the relevant pathology and related information, the interaction with the Hospital Information System requires dedicated personnel and an in-depth knowledge of the IT architecture of the Hospital. In this paper, we show an ad-hoc solution for the RE-SAMPLE project in Fondazione Policlinico Universitario Agostino Gemelli IRCCS, where the Chronic Obstructive Pulmonary Disease (COPD) is studied and a framework for managing that pathology is proposed. The final aim of this work is to provide a description of the tailored procedures of data extraction, integration and harmonization, and the final creation of a dedicated COPD data mart for research purposes that has been implemented in the hospital premises by Gemelli Generator RWD R&D.
Download
Paper Nr: 114
Title:

Analysis of the Relationship Between Intelligence, Sensory Processing Sensitivity and the Digital Tree Drawing Test: A Feasibility Study
Authors:

Johanna Emelie Heger, Dorothea Isselstein-Mohr, Sebastian Unger and Thomas Ostermann
Abstract: The personality trait of intelligence has a research history rich in psychometric tradition, whereas sensory processing sensitivity is a young construct, which in its conceptualization shows similarities with other psychological and psychopathological concepts such as introversion, autism spectrum disorder, but also various giftedness concepts. The digital tree drawing test recently achieved good results in the diagnostics of cognitive performance losses in adults. The present study investigates whether the characteristics of intelligence and sensitivity are related and can be mapped in a second step using the digital tree test in the drawing process. For this purpose, 19 children and adolescents with existing intelligence and sensitivity diagnoses underwent the digital tree test. The results were evaluated using correlation analyses. Hardly any significant correlations were found between intelligence and sensitivity. Contrary to the previous assumption, the correlations found were negative. Drawing parameters, on the other hand, showed clear correlations with both traits, but here primarily with the sensitivity facets, so that drawing process variables could be identified which appear to be relevant for the personality traits. Future research could investigate in greater depth the direction and predictive value of these correlations in order to expand the diagnostic repertoire of psychological practitioners using the digital tree drawing test.
Download
Paper Nr: 115
Title:

The Development and Validation of the “Attitudes Towards Digitalization” (Att-Dig) Questionnaire
Authors:

Daniel Niewrzol, Jan P. Ehlers and Thomas Ostermann
Abstract: Attitudes towards digitalization play a major role in almost all areas of human interaction including the health care system. Unfortunately, existing assessments and respective instruments on attitudes towards digitalization are often negatively framed, while balanced and broader approaches exist only marginally. The aim of this work was therefore to develop an assessment instrument from a self-generated item pool capturing a broad range of aspects of attitudes towards digitalization. Items were answered in an online survey by a total of 214 participants (mean age: 30.8±14.4 years 56,1% female). A principal component analysis was performed and 5 subscales “Digitalisation and Social Life” (5 items, Cronbach's alpha=0.789),” Digitalisation and Loss of Control” (4 items, Cronbach's alpha=0.817), ” Digitalisation, Knowledge and Education” (4 items, Cronbach's alpha=0.791), ” Digitalisation and Gain of freedom” (3 items, Cronbach's alpha=0.749), and ” Digitalisation, Equity and Prosperity” (3 items, Cronbach's alpha=0.699) were extracted covering 63.5% of the item variance, showing a sufficient internal consistency of the subscales. There were significant differences for some of the subscales with regard to gender, age, and education. Only weak and non-significant correlations were found with respect to the subscales “self-efficacy”, “optimism”, and “pessimism” of the SWOP-K9 questionnaire. Thus, in sum, although there is a need for further research, the Att-Dig is a sound survey instrument to economically assess the attitude towards digitalisation. It can be used in different areas of public life and health care and is easy and quick to answer.
Download
Paper Nr: 135
Title:

Performance Analysis of a Data Stream Processing System for Online Activity Classification via Wearable Sensor Data
Authors:

Hawzhin Hozhabr Pour, Gabriela Ciortuz, André Lüers and Sebastian Fudickar
Abstract: Online activity recognition based on wearable sensors is commonly used in sports and medicine applications. The question of whether cloud or edge computing approaches are more suitable is not easy to answer and depends on several factors. To address this issue, the influence the resource availability, batch sizes and number of considered users on the throughput and latency of central data stream processing architectures has yet to be answered. This article conducts a performance analysis, identifying relevant factors for a corresponding cloud-based online data stream processing platform for online human activity recognition, using the Apache Spark data processing framework and the Apache Kafka distributed messaging system. The platform focuses on quantitative performance criteria to evaluate its effectiveness in terms of latency (turnaround time) and throughput (number of users). Both metrics, throughput and latency (dependent variables), depend on the batch interval, number of users, and hardware availability (independent variables). In addition to identifying clear advantages of larger batch intervals, we also found significant benefits in applying vertical scaling. The results indicate a monthly cost of 1e per user for compute resources in online activity recognition, a price that could potentially be reduced by combining edge and cloud computing.
Download
Paper Nr: 164
Title:

Comprehensive Evaluation of Regression and Classification Models on Brain Stroke Datasets
Authors:

Dimitar Trajkov, Ana Kostovska, Panče Panov and Dragi Kocev
Abstract: This paper investigates the application of machine learning models for predicting brain stroke outcomes, leveraging publicly available datasets. We evaluate the performance of various classification and regression models, including ensemble methods such as AdaBoost, Gradient Boosting, and Random Forest, across eight datasets related to stroke prediction. Our results show that data quality and dataset characteristics have a more significant impact on model performance than the choice of algorithm, underscoring the importance of high-quality, well-curated data in achieving accurate and reliable predictions. Additionally, we emphasize the need for transparency, reproducibility, and traceability in AI research, highlighting the challenges associated with the scarcity of publicly available stroke datasets. This study provides a foundation for developing more trustworthy AI tools for stroke prediction and encourages further efforts in data sharing and model validation.
Download
Paper Nr: 176
Title:

Assessing the Practicality of Designing a Comprehensive Intelligent Conversation Agent to Assist in Dementia Care
Authors:

Ritwik Raj Saxena and Arshia Khan
Abstract: Tools and techniques powered by artificial intelligence (AI) and its subfields like machine learning (ML) and natural language processing (NLP) have pervaded most disciplines across the global technological, economical, and sociocultural landscapes. In most areas, the permeation of AI has shown exceptional promise. Medicine and healthcare constitute a domain which has not remained aloof from the positive implications of harnessing AI. AI-driven tools, for instance in neuroimaging and health monitoring, have painted a tapestry of encouraging possibilities in this province. Such tools have found application in fields like assisting diagnosis, disease progression tracking, and patient management in many subjects within medicine. Intelligent conversation agents, more informally referred to as AI-based chatbots, form one of the most prevalent applications of AI. AI-fueled chatbots like ChatGPT have made rampant inroads into the lives of countless people around the world, easing innumerable routine tasks they are responsible for. This article offers a systematic but succinct overview of dementia, and, in this backdrop, explores the potential efficacy of a proposed intelligent conversation agent aimed at sufficing the fulfilment of the care-associated requirements of various stakeholders in dementia care. We provide an outline and a critical assessment and suggest future directions on the adoption of such a tool. We conclude that a smart conversation agent has the potential to positively overhaul the extant worldwide paradigm of dementia care.
Download
Paper Nr: 182
Title:

Integrating Clinical Expertise into Software Development: Evaluating the Use of openEHR Archetypes for Requirements Elicitation in Healthcare Applications
Authors:

José Silva and André Araújo
Abstract: OpenEHR archetypes are standardized frameworks designed to model clinical information in healthcare systems, enabling a consistent and integrated representation of complex data. These models encompass common clinical elements such as symptoms, diagnoses, and treatments, ensuring that information is understood and applied uniformly across different contexts. This paper investigates the use of openEHR archetypes in the requirements elicitation and specification process for clinical systems, highlighting their potential to integrate healthcare professionals’ knowledge into software development. The literature review reveals a significant gap in the participation of these professionals during the requirements elicitation phase, especially in studies that apply archetypes. Quantitative and qualitative results positively perceive the methodology used, highlighting clarity, collaboration, and alignment with end-user needs. Statistical analysis using the Wilcoxon test presented significant p-values, indicating that professionals considered the method straightforward, intuitive, and conducive to engagement, with real opportunities for contribution to the validation of requirements. The qualitative data reinforce the importance of a collaborative environment and suggest the need for deeper involvement of healthcare professionals at all process stages. In conclusion, this research indicates that applying openEHR archetypes, combined with more significant interaction with healthcare professionals, is promising for integrating clinical expertise effectively and directly into developing clinical systems.
Download
Paper Nr: 190
Title:

A Proposal for Explainable Breast Cancer Detection from Histological Images
Authors:

Lucia Lombardi, Myriam Giusy Tibaldi, Rachele Catalano, Mario Cesarelli, Antonella Santone and Francesco Mercaldo
Abstract: Breast cancer is the most prevalent cancer among women globally, making early and accurate detection essential for effective treatment and improved survival rates. This is the reason why, early and accurate breast cancer detection is crucial for proper treatment planning to save a life. This paper presents a method designed to detect and localize breast cancer using deep learning, specifically convolutional neural networks. The approach classifies histological images of breast tissue as either tumor-positive or tumor-negative. We utilize several deep learning models, including a custom-built CNN, EfficientNet, ResNet50, VGG-16, VGG-19, and MobileNet. Fine-tuning was also applied to VGG-16, VGG-19, and Mo bileNet to enhance performance. The aim is to provide a more effective network, able to correctly detect and localise breast cancer, that could support the physician in making clinical decisions. It could also prove to be a successful model to speed up the diagnostic process and detect the possible presence of the disease at an early stage. Additionally, we introduce a novel deep learning model called MR Net, aimed at providing a more accurate network for breast cancer detection and localization, potentially assisting clinicians in making informed decisions. This model could also accelerate the diagnostic process, enabling early detection of the disease. Furthermore, we propose a method for explainable predictions by generating heatmaps that highlight the regions within tissue images that the model focuses on when predicting a label, revealing the detection of benign, atypical, and malignant tumors. We evaluate both the quantitative and qualitative performance of MR Net and the other models, also presenting explainable results that allow visualization of the tissue areas identified by the model as relevant to the presence of breast cancer.
Download
Paper Nr: 192
Title:

Multi-Modal Framework for Autism Severity Assessment Using Spatio-Temporal Graph Transformers
Authors:

Kush Gupta, Amir Aly and Emmanuel Ifeachor
Abstract: Diagnosing Autism Spectrum Disorder (ASD) remains challenging, as it often relies on subjective evaluations and traditional methods using fMRI data. This paper proposes an innovative multi-modal framework that leverages spatiotemporal graph transformers to assess ASD severity using skeletal and optical flow data from the MMASD dataset. Our approach captures movement synchronization between children with ASD and therapists during play therapy interventions. The framework integrates a spatial encoder, a temporal transformer, and an I3D network for comprehensive motion analysis. Through this multi-modal approach, we aim to deliver reliable ASD severity scores, enhancing diagnostic accuracy and offering a scalable, robust alternative to traditional techniques.
Download
Paper Nr: 201
Title:

Evidence on Robotic Prostatectomy: Discussing the Limitations of Real-World Data
Authors:

Maciej Dzik, Kacper Mucha and Monika Zaleska
Abstract: INTRODUCTION: The aim of this research is to evaluate Real-World Data (RWD) related to prostatectomy outcomes for prostate cancer with a focus on identifying potential biases and data limitations. METHODS: This study was based on the financial records collected in the database the Polish National Health Fund. The sample included 14,376 patients who underwent robot-assisted radical prostatectomy (RARP), laparoscopic radical prostatectomy (LRP) or conventional radical prostatectomy (CRP) between 20 September 2022 and 31 December 2023. Comparative analysis focused exclusively on the duration of hospitalisation. Additional outcomes included mortality. RESULTS: In total 6,609 patients had RARP. RARP compared to both CRP and LRP was associated with a reduction in inpatient days by 2.81 (95% CI: -2.98, -2.65; p<0.0001) and 0.91 (95% CI: -1.02, -0.8; p<0.0001) respectively. Patient admitted as emergencies had statistically longer hospital stays by 1.03 days (p<0.0001). CONCLUSIONS: The overall length of hospitalization has been reduced, but interpreting the results obtained from RWD in terms of relative benefits is challenging. The analysis faced several challenges, including interpreting outcome measures and validating their clinical significance, handling outliers, addressing non-random assignment, and accounting for unobserved covariates. These limitations underscore the need for further research to enhance the quality of comparisons.
Download
Paper Nr: 217
Title:

Coastal and Rural Digital Exclusion: The Case for Voice AI
Authors:

Rory Baxter, Oksana Hagen, Amir Aly, Ray B. Jones and Katharine Willis
Abstract: The adoption of digital platforms for health and social resources disadvantages vulnerable populations, including older adults and those living in remote or deprived areas. This could be remediated using voice-based conversational AI (Voice AI) systems delivered via landline phone, bypassing the requirement for digital device access or digital skills. British rural and coastal regions often have poorer digital infrastructure and pockets of deprivation, and consequently higher levels of digital exclusion. This study explores digital exclusion in Southwest England and the suitability of Voice AI systems for supporting digital inclusion. Seventeen participants aged 50 years or over were interviewed by telephone and took part in one of two face to face focus groups to identify how digital exclusion impacts access to health and wellbeing resources. The results indicated that digital access was severely impacted by unreliable infrastructure and exacerbated by limited digital skills. Phone-based Voice AI systems could then provide viable solutions to support access to digital health and social resources for digitally marginalised coastal and rural communities.
Download
Paper Nr: 231
Title:

Approaches to Promoting Patients’ and Citizens’ eHealth Literacy
Authors:

Sophia Grummt and Brita Sedlmayr
Abstract: eHealth Literacy (eHL) is of crucial importance in the increasingly digital landscape of healthcare. eHL is defined as the intersection of general health literacy and digital competencies, expanded to include new facets such as data literacy and privacy awareness. Current studies indicate that a significant portion of the German population has low eHL, leading to difficulties in evaluating online health information and challenges in accessing and utilizing digital health services, which correlate with lower health status. Various measures for enhancing citizens’ eHL are proposed, including education, public awareness campaigns, and ensuring equal access to digital health services. Specific initiatives from the "MiHUBx" project are highlighted, such as developing a knowledge platform for patients and organizing information events focused on digital health topics. Fostering eHL is a societal responsibility that requires an inclusive and coordinated approach. Strategic efforts to this end are vital to ensure that all citizens can benefit from advancements in digital healthcare.
Download
Paper Nr: 238
Title:

Predicting Falls from Operational Data: Insights and Limitations of Using a Non-Specialized Database
Authors:

Julien Räker, Patrick Elfert, Cletus Brauer, Marco Eichelberg, Frerk Müller-von Aschwege and Andreas Hein
Abstract: Falls among the elderly are a significant public health concern. This study investigates the feasibility of predicting falls using an operational dataset from Johanniter-Unfall-Hilfe (JUH) home emergency call system, which was not created under laboratory conditions for scientific purposes. An anonymized dataset containing records from 160,281 participants in Germany was analyzed. Statistical analysis identified 104 out of 400 features significantly associated with falls, though with weak correlations (Cramer’s V ranging from 0.006 to 0.071). A one-class Support Vector Machine (SVM) was employed due to the absence of explicit non-fall cases, achieving a true positive rate of 55.10%. The lack of explicit non-fall data prevented evaluation of specificity and overall accuracy. The study demonstrates the potential of using operational datasets for fall prediction but highlights significant limitations due to data quality issues, such as the lack of explicit fall records, absence of non-fall cases, lack of temporal data, and missing values. Recommendations are made to improve data collection practices to enhance the utility of such datasets for predictive modeling.
Download
Paper Nr: 251
Title:

A Pre-Study on Tremor Classification During Activities of Daily Living
Authors:

Linda Büker, Sandra Hellmers and Andreas Hein
Abstract: Motor impairments, such as tremors, are often measured with specific tests or rating scales. As these have some disadvantages, like an inter-rater reliability and a lack of representation of the everyday life, a sensor-based continuous and objective monitoring of activities of daily living could be a suitable alternative. According to the literature, the use of inertial measurement units attached to the tremor-dominant arm in combination with support vector machines or neural networks seem to be promising. However, many approaches have to be adapted individually. Therefore, we conducted a preliminary study with ten healthy participants, who were asked to perform conventional and simulated tremor movements during five different activities related to eating. These movements were recorded with inertial measurement units. We identified four different parameters calculated from the recorded data, that we used to train multiple support vector machines for a non-individualized approach. The overall median accuracy score was 0.75, which is comparable to the results reported in the literature. This shows that support vector machines may be a non-individualized approach for differentiating between tremor and non-tremor movements during activities of daily living.
Download
Paper Nr: 265
Title:

Predicting Adverse Events in Developmental Disabilities Population
Authors:

James P. McGlothlin, Micah Price-Offerman, Robbie Beyer, George Casey and John P. Barile
Abstract: Individuals with development disabilities can experience a variety of adverse events. We have found that these events are often unreported. In this project, we work with a large government program which assists such individuals. The goal of the project is to use artificial intelligence (AI) and other modern technologies to predict adverse events. This will allow case managers to better avoid adverse events, prepare for them and help the program participants. Our initial results show very good accuracy and precision in identifying risk and predicting participant adverse events.
Download
Paper Nr: 281
Title:

MIDTs: Interdisciplinary Method for Technological Research Development with a Focus on Health
Authors:

José Eurico de Vasconcelos Filho, Joel Sotero da Cunha Neto and José Fernando Ferreira
Abstract: This study introduces MIDTs, an interdisciplinary method for technological research aimed at healthcare applications. MIDTs integrates Design Science and User-Centered Design principles, structured into six phases to ensure both scientific rigor and practical applicability. Over the last decade, it has been applied to more than 70 projects, generating academic theses, patents, and clinical solutions. Empirical evidence indicates that MIDTs fosters innovative and user-centered outcomes, effectively addressing complex societal demands within healthcare. The method’s adaptability is further demonstrated through its potential application in other sectors. By providing a clear framework for interdisciplinary collaboration and solution development, MIDTs offers a robust approach to bridge research, practice, and user needs in technology-driven health initiatives.
Download
Paper Nr: 290
Title:

Preliminary Usability Evaluation of a Virtual Reality (VR) Application for Quitting Nicotine Vaping
Authors:

Bethany K. Bracken, Phillip C. Desrochers, Ian McAbee, Nicolette M. McGeorge, Susan Latiff, Bradly T. Stone, Dan T. Duggan, Corinne Cather and A. Eden Evins
Abstract: Nicotine vaping is a global problem. Limited vaping cessation interventions are available; and current treatments have limited accessibility due to systemic barriers to care (e.g., scarcity of treaters). Digital therapeutics (DTx) can reduce these barriers. We have embedded standard cognitive behavioral therapy (CBT) content into virtual reality (VR) to create a VR-based app focused on vaping cessation: Novel, On-demand VR for Accessible, Practical, and Engaging therapy (NO VAPE). NO VAPE allows users to practice CBT skills gained in traditional therapy through an accessible, immersive, and engaging platform. Our ultimate goal is to conduct a full clinical trial to test whether NO VAPE motivates greater intervention adherence and satisfaction. To prepare, we conducted a usability study with N = 6 young adults who currently vape, aiming to evaluate safety, usability, and overall enjoyment of NO VAPE. We categorized errors into categories in ascending severity from minor usability errors to safety violations. There were no safety violations by any participants providing evidence that the app is low-risk and safe (from a software use perspective, not a substance use perspective). Participant reported high levels of enjoyment, said they would like to use NO VAPE again, and did not experience symptoms of simulator sickness. We also identified multiple software bugs we are now addressing.
Download
Paper Nr: 304
Title:

Protocol Design for in-Class Projects: Comparative Analysis of EEG Signals Among Sexes
Authors:

Shujauddin Syed, Sifat Redwan Wahid, Ted Pedersen, Jack Quigley and Arshia Khan
Abstract: This paper focuses on the development of a structured protocol to support undergraduate students in conducting in-class projects. Project-Based Learning (PBL) has gained recognition as an effective educational approach, offering students practical, hands-on experience and fostering a deeper understanding of the application of theoretical concepts. Despite its advantages, undergraduate students often face challenges in successfully completing in-class projects due to the lack of well-defined protocols to guide their efforts. To address this gap, we, a team of graduate students serving as teaching assistants (TAs), designed this protocol based on their close interaction with undergraduates and an understanding of the challenges they face. This protocol aims to enhance the ability of undergraduate students to complete their project in a systematic and structured way. To demonstrate the implementation, we provide a step-by-step guide based on an in-class project conducted as part of the “Sensors and IoT” course (CS4432/5432) at the University of Minnesota Duluth.
Download
Paper Nr: 310
Title:

Key Factors in Achieving Intersectoral Interoperability: A Scoping Review
Authors:

Eveline Prochaska, Franziska Bathelt, Michelé Zoch, Antonia Ewald and Elisa Henke
Abstract: Intersectoral interoperability is a fundamental basis for effective collaboration and seamless information exchange across various sectors of the healthcare system. This paper presents a scoping review to examine the current state of research into intersectoral interoperability, focusing on the technical, syntactic, semantic, and organizational levels. Key factors identified include the adoption of international standards for data formats, terminologies, and communication protocols, as well as the establishment of trusted governance structures and compliance with ethical and legal requirements. Syntactic interoperability was most frequently addressed, followed by technical and semantic aspects, with organizational factors also playing a significant role.
Download
Paper Nr: 311
Title:

Anxiety Detection in Reddit Posts Through Emotion Dynamics Analysis
Authors:

Ashala Senanayake and Zilu Liang
Abstract: Anxiety disorders have become a significant subset of mental health challenges in the context of complex modern social life. The widespread integration of social media into daily life has created platforms for individuals to share their updates, offering a rich resource for linguistic and behavioural analysis outside traditional clinical settings. Among these platforms, Reddit stands out as a valuable tool for researchers due to its rich and diverse textual data. This paper leverages five commonly used machine learning models: decision tree, random forest (RF), k-nearest neighbours, linear regression, and naive Bayes to explore the emotional dynamics present in Reddit posts for detecting anxiety. Reddit posts from 1,800 users, categorized as either anxiety or non-anxiety, were used for model training, validation, and testing, with data split into 70%, 15%, and 15%, respectively. The decision-making process of the best-performing model was evaluated by incorporating feature importance. RF achieved the best performance among all models, with an accuracy of 89%. Its interpretation revealed that average emotion scores and normalized emotion gaps were key factors, highlighting the significance of emotional intensity and variability over time. Furthermore, the results indicate that emotions such as sadness and joy play a particularly significant role in detecting anxiety.
Download
Paper Nr: 320
Title:

Design Thinking Process for a Gamified Mobile App to Improve Migrants’ Well-Being and Inclusion
Authors:

Francesca Pia Travisani, Veronica Rossano and Enrichetta Gentile
Abstract: An app with digital storytelling and gamification represents a key element in promoting the learning of the typical terminology of the socio-health context that aims to ensure the psychophysical well-being and inclusion of migrants by removing language and cultural barriers, improving access to essential Italian health and social services by using sports as a vehicle for learning and socialization. The application could be integrated as an educational tool in integration centres or schools to encourage understanding of health concepts in an interactive and engaging way. The research study conducted the iterative user-centred “Design Thinking Process”, which includes the phases of empathy, definition, ideation, prototyping, and testing. In the living labs, we have gathered feedback from ten adolescent migrants who were co-designers of the iterative development, which enabled us to collect data to evaluate the app's usability, effectiveness, and social impact on migrant quality of life. Analysis of the feedback revealed that the app's usability and intuitiveness have the potential to be effective and well-accepted, as they enable language skills about social and health terms simply and pleasantly and consequently facilitate access to social and health services by improving migrants' mental-physical health and social integration.
Download