HEALTHINF 2022 Abstracts

Full Papers Short Papers
Full Papers
Paper Nr: 6
Title:

Factors Influencing Tele-dermatology Adoption among the Lebanese Youth: A Pilot Study at Saint Joseph University
Authors:

Nabil G. Badr, Nanor Aroutine and Joumana Yeretzian
Abstract: The demand and use for Tele-dermatology (TDM) to diagnose skin lesions is rising worldwide. Using the technology acceptance model, we evaluate the factors influencing the acceptance of Tele-Dermatology to diagnose skin lesions among the Lebanese students. We complete a pilot study with Lebanese students from Saint Joseph University of Beirut (USJ). After examining the responses in a descriptive analysis, we develop some initial hypotheses and proceed to build the statistical model to test them using Smart PLS3. Our findings show that 64% of the students are ready to use Tele-Dermatology in their everyday life. Most of those students are females between 18 and 24 years old. Wrapping up our results, information from this study indicates that marital status is most likely a determinant of intention to use TDM among students – whereby, most single students are ready now (65%) while most married students are inclined to use it in the future (67%). The study also suggests that the Lebanese youth prioritize result demonstrability as a factor in their intention to use TDM. Further, mobile TDM must save them time must be easy to use to be perceived useful.
Download
Paper Nr: 11
Title:

Forecasting Thresholds Alarms in Medical Patient Monitors using Time Series Models
Authors:

Jonas Chromik, Bjarne Pfitzner, Nina Ihde, Marius Michaelis, Denise Schmidt, Sophie I. Klopfenstein, Akira-Sebastian Poncette, Felix Balzer and Bert Arnrich
Abstract: Too many alarms are a persistent problem in today’s intensive care medicine leading to alarm desensitisation and alarm fatigue. This puts patients and staff at risk. We propose a forecasting strategy for threshold alarms in patient monitors in order to replace alarms that are actionable right now with scheduled tasks in an attempt to remove the urgency from the situation. Therefore, we employ both statistical and machine learning models for time series forecasting and apply these models to vital parameter data such as blood pressure, heart rate, and oxygen saturation. The results are promising, although impaired by low and non-constant sampling frequencies of the time series data in use. The combination of a GRU model with medium-resampled data shows the best performance for most types of alarms. However, higher time resolution and constant sampling frequencies are needed in order to meaningfully evaluate our approach.
Download
Paper Nr: 12
Title:

A Novel Atomic Annotator for Quality Assurance of Biomedical Ontologies
Authors:

Rashmi Burse, Michela Bertolotto and Gavin Mcardle
Abstract: Existing lexical auditing techniques for Quality Assurance (QA) of biomedical ontologies exclusively consider lexical patterns of concept names and do not take semantic domains associated with the tokens constituting those patterns into consideration. For many similar lexical patterns the corresponding semantic domains may not be similar. Therefore, not considering the semantic aspect of similar lexical patterns can lead to poor QA of biomedical ontologies. Semantic domain association can be accomplished by using a Biomedical Named Entity Recognition (Bio-NER) system. However, the existing Bio-NER systems are developed with the goal of extracting information from natural language text, like discharge summaries, and as a result do not annotate individual tokens of a clinical concept. Annotating individual tokens of a clinical concept with their semantic domains is important from a QA perspective, since these annotations can be leveraged to gain insight into the type of attributes that should be associated with the concept. In this paper we present an annotator that atomically annotates the tokens of a clinical concept by crafting atomic dictionaries from the sub-hierarchies of Systematized Nomenclature of Medicine (SNOMED). Semantic analysis of lexically similar concepts by atomically annotating semantic domains to the tokens will ensure improved QA of biomedical ontologies.
Download
Paper Nr: 18
Title:

Ten Years of eHealth Discussions on Stack Overflow
Authors:

Pedro M. Oliveira, Evilasio Costa Junior, Rossana C. Andrade, Ismayle S. Santos and Pedro S. Neto
Abstract: Over the past decade, we have seen growth in the usage of technologies in health. However, few papers are addressing the perspective reported by practitioners during the development of healthcare solutions. This perspective is relevant to identifying the most used strategies in this area and what challenges persist. Thus, this work analyzed eHealth discussions from Stack Overflow (SO) to understand the eHealth developers’ behavior. Using a KDD-based process, we got and processed 6,082 eHealth questions. The most discussed topics include manipulating medical images, electronic health records with the HL7 standard, and frameworks to support mobile health (mHealth) development. Concerning the challenges faced by these developers, there is a lack of understanding about the DICOM and HL7 standards, the absence of data repositories for testing, and the monitoring of health data in the background using mobile and wearable devices. Our results also indicate that discussions have grown mainly on mHealth, primarily due to monitoring health data through wearables.
Download
Paper Nr: 19
Title:

Best Practice in Multi-organisation Sensitive Health Data Sharing: A Comparative Analysis of Ireland’s Data Governance Approach for the Covid–19 Data Research Hub
Authors:

Aleksandra Czarnik, Aoife Darragh, Maria Hurley, Daniel O’Connell, Michele Quagliata and Rob Brennan
Abstract: This paper examines, from a data governance perspective, the creation and operation of the Irish Covid–19 Data Research Hub, a secure multi-institution collation and access-controlled source of sensitive Covid–19 epidemiological data from diverse sources. The Hub is assessed alongside international comparators and with reference to a set of leading academic data governance models, including those developed by Khatri & Brown (2010), Winter & Davidson (2019), and Abraham et al (2019). The analysis explores the requirements for such data hubs balancing data protection, security, and health policy decision making. It examines the data hub design from architectural, data access policy, and data governance perspectives. Whilst recognising certain unique features of the Covid–19 Data Research Hub not replicated elsewhere, it highlights key data governance strengths and gaps in the model used which may inform future development of similar hubs supporting the exploitation of public sector data for health policy-related research. The interdisciplinary legal and technical data governance assessment methodology described here is applicable to the increasing number of data federation and aggregation projects increasingly being deployed in both public and private healthcare settings.
Download
Paper Nr: 31
Title:

A Robust Approach for a Real-time Accurate Screening of ST Segment Anomalies
Authors:

Giovanni Rosa, Marco Russodivito, Gennaro Laudato, Angela R. Colavita, Simone Scalabrino and Rocco Oliveto
Abstract: Nowadays, Computerized Decision Support Systems (CDSS) play an important role in medical support and preventative care. In those scenarios, the monitoring of biomedical data, such as the ECG signal, is fundamental. The ECG signal may reveal a variety of abnormalities or pathological conditions. Some examples are Ischemia and Myocardial Infarction (MI), with a significant impact on the world’s population. Both these conditions can be diagnosed by observing changes in specific sections of the ECG, such as the ST segment and/or T-wave of heartbeats. Much effort was devoted by the scientific community to aim at automatically identifying ST anomalies. The main drawback of such approaches is often a trade-off between the accuracy in the classification, the robustness to noise, and the real-time responsiveness. In this work, we present RAST, a robust approach for a Real-time Accurate screening of ST segment anomalies. RAST takes as input a sequence of 10 successive heartbeats extracted from an ECG recording and provides as output the classification of the ST segment trend. We evaluated two versions of RAST, namely RAST-BINARY, and RAST-TERNARY: the first capable of distinguishing only between an ST anomaly and Normal Sinus Rhythm and the second able to distinguishing between ST elevation, ST depression, and normal rhythm. Moreover, we conducted an extensive study by experiment also (i) the validation within the intra- and inter-patient strategies and (ii) the ideal number of successive heartbeats in which to observe an anomalous episode of change in the ST segment. As a result, both RAST-BINARY and RAST-TERNARY can achieve an F1 score of 0.94 with a window of 4 heartbeats in the inter-patient validation. For the intra-patient validation, both versions achieve an F1 score of 0.73 using a longer observation window.
Download
Paper Nr: 36
Title:

Value-based Consent Model: A Design Thinking Approach for Enabling Informed Consent in Medical Data Research
Authors:

Simon Geller, Sebastian Müller, Simon Scheider, Christiane Woopen and Sven Meister
Abstract: Due to new technological innovations, the increase in lifestyle products, and the digitalisation of healthcare the volume of personal health data is constantly growing. However, in order to use, re-use, and link personalised health data and, thus, unlock their potential benefits in health research, the authors of the data need to voluntarily give their informed consent. That is a major challenge to health data research, because the classic informed consent process requires the immense administrative burden to ask for consent, every time personal health data is accessed. In this paper we argue that all alternative consent models that have been developed to tackle this problem, either do not reduce administrative burdens significantly or do not conform to the informed consent ideal. That is why we used the design thinking approach to develop an alternative consent model that we call the value-based consent model. This model has the potential to reduce administrative burdens while empowering research subjects to autonomously translate their values into consent decisions.
Download
Paper Nr: 46
Title:

Data Balancing using Deep Convolutional Generative Adversarial Networks (DCGAN) in Patients with Congenital Syndrome by Zika Virus
Authors:

Érika G. Assis, Mark A. Song, Luis E. Zárate and Cristiane N. Nobre
Abstract: Class imbalance is a common health care problem and often affects the performance of machine learning algorithms. Unfortunately, the minority class, generally the one with the most significant interest, has their learning affected to the detriment of the majority class. This article proposes using Deep Convolutional Generative Adversarial Networks (DCGAN) for minority class oversampling, generating synthetic instances. For this, the ’RESP-Microcephaly’ database was used, which records suspected cases of congenital alteration due to Zika virus (ZIKV) infection. The database presents unbalanced data with 2904 and 7606 instances with and without congenital alteration, respectively. To evaluate the performance of DCGAN, we compared this method with an undersampling and an oversampling approach, using SMOTE with three classification algorithms. The use of DCGAN for balancing demonstrates a significant improvement in classification indices, especially about the minority class.
Download
Paper Nr: 47
Title:

Selection of Representative Instances using Ant Colony: A Case Study in a Database of Children and Adolescents with Attention-Deficit/Hyperactivity Disorder
Authors:

Henrique R. Hott, Caroline R. Jandre, Pedro S. Xavier, Amal Miloud-Aouidate, Débora M. Miranda, Mark A. Song, Luis E. Zárate and Cristiane N. Nobre
Abstract: Instance Selection (IS) helps select the most notable instances from the database, improving its characterization and relevance. In this context, this article applies the IS, using the Ant Colony Optimization (ACO) heuristic, to obtain more efficient classification models in the identification of school performance, in arithmetic, writing, and reading, of children and adolescents with Attention-Deficit/Hyperactivity Disorder (ADHD), characterized by excessive symptoms of inattention, hyperactivity, and impulsivity. The Random Forest, Neural Networks, KNN, and CART classifiers were used to evaluate the performance of the selection performed by the ACO method. With the ACO, it was possible to obtain a gain of 20 percentage points with the KNN (K = 1), in arithmetic, in the metric F-measure, referring to the upper class, the minority class. The results achieved show the excellent efficiency of the ACO in this study.
Download
Paper Nr: 48
Title:

A Framework for AI-enabled Proactive mHealth with Automated Decision-making for a User’s Context
Authors:

Muhammad Sulaiman, Anne Håkansson and Randi Karlsen
Abstract: Health promotion is to enable people to take control over their health. Digital health with mHealth empowers users to establish proactive health, ubiquitously. The users shall have increased control over their health to improve their life by being proactive. To develop proactive health with the principles of prediction, prevention, and ubiquitous health, artificial intelligence with mHealth can play a pivotal role. There are various challenges for establishing proactive mHealth. For example, the system must be adaptive and provide timely interventions by considering the uniqueness of the user. The context of the user is also highly relevant for proactive mHealth. The context provides parameters as input along with information to formulate the current state of the user. Automated decision-making is significant with user-level decision-making as it enables decisions to promote well-being by technological means without human involvement. This paper presents a design framework of AI-enabled proactive mHealth that includes automated decision-making with predictive analytics, Just-in-time adaptive interventions and a P5 approach to mHealth. The significance of user-level decision-making for automated decision-making is presented. Furthermore, the paper provides a holistic view of the user's context with profile and characteristics. The paper also discusses the need for multiple parameters as inputs, and the identification of sources e.g., wearables, sensors, and other resources, with the challenges in the implementation of the framework. Finally, a proof-of-concept based on the framework provides design and implementation steps, architecture, goals, and feedback process. The framework shall provide the basis for the further development of AI-enabled proactive mHealth.
Download
Paper Nr: 52
Title:

A Digital, Game-based Application to Support Treatment of Parkinson’s Disease: A Design Thinking Approach
Authors:

Anne Mainz and Sven Meister
Abstract: One of the most common neurodegenerative disorders that affects more and more people at an advanced age is Parkinson’s disease. Patients suffer from various symptoms and especially the motor restrictions and psychological symptoms worsen the quality of life of the affected persons. The physical therapy for this disease to improve motor performance and complementary exercises is characterised by repetitive training and patients often suffer from a strong exhaustion and lack of motivation due to their disease. To address these problems, a serious game concept for Parkinson's therapy was developed. The concept was created using the Design Thinking methodology for a user-centred design. The final result is the concept and prototype of a competitive multiplayer exergame that was developed to increase the motivation of the patients to participate through social play and the idea of competition in order to support the motor therapy of Parkinson’s disease patients.
Download
Paper Nr: 55
Title:

Adversarial Evasion Attacks to Deep Neural Networks in ECR Models
Authors:

Shota Nemoto, Subhash Rajapaksha and Despoina Perouli
Abstract: Evasion attacks produce adversarial examples by adding human imperceptible perturbations and causing a machine learning model to label the input incorrectly. These black box attacks do not require knowledge of the internal workings of the model or access to inputs. Although such adversarial attacks have been shown to be successful in image classification problems, they have not been adequately explored in health care models. In this paper, we produce adversarial examples based on successful algorithms in the literature and attack a deep neural network that classifies heart rhythms in electrocardiograms (ECGs). Several batches of adversarial examples were produced, with each batch having a different limit on the number of queries. The adversarial ECGs with the median distance to their original counterparts were found to have slight but noticeable perturbations when compared side-by-side with the original. However, the adversarial ECGs with the minimum distance in the batches were practically indistinguishable from the originals.
Download
Paper Nr: 56
Title:

Improved Blood Vessels Segmentation of Retinal Image of Infants
Authors:

Vijay Kumar, Het Patel, Kolin Paul, Abhidnya Surve, Shorya Azad and Rohan Chawla
Abstract: Retinopathy of prematurity (ROP) is the leading cause of blindness in premature babies worldwide. ROP is quantified through the structural information of the retinal vessel map, such as vessels width, tortuosity and extent. Therefore, the accuracy of quantitative studies depends on the quality of the segmented blood vessels’ map. Fundus images used for neonatal eye examination are prone to many artefacts and noises due to patient movement, erratic illumination, improperly focused camera sensor, etc. Existing vessel segmentation algorithms work well on retinal images of adults but fail to detect underdeveloped vessel structures of neonatal’s fundus images. At the same time, the unavailability of fundus images of infants has hindered the development of the data-driven methods for vessel segmentation. This work proposes a new Deep Convolutional Neural Network (DCNN) based vessels segmentation system for the screening for the neonatal eye disorder ROP. The proposed system uses a DCNN, Generative Adversarial Network (GAN) Pix2Pix or U-Net for vessel segmentation. Using publicly available fundus image datasets, we used an efficient and robust training procedure for the proposed system and tested them with preterm neonatal eye images from a local hospital. Experimental results show that the proposed system allows better screening of ROP with robustness to noises and inter-class variation. It has achieved an average accuracy of 96.69% for vessels segmentation and a Dice coefficient between 0.60 and 0.64. Our system is able to achieve an accuracy of 88.23% for Zone-1 case of ROP.
Download
Paper Nr: 58
Title:

Classification Models for Predicting Inflammatory Bowel Disease Healthcare Utilization
Authors:

Dmitriy Babichenko, Behnam Rahdari, Ben Stein, Suraj Subramanian, Claudia R. Rivers, Gong Tang and David Binion
Abstract: Objective. Inflammatory Bowel Disorders (IBD) is a group of gastric disorders that include well-known maladies such as Crohn’s disease and Ulcerative Colitis (UC), as well as a number of other gastric disorders that are not well classified. Subgroups of patients contribute disproportionately to treatment costs. This work aims to create and evaluate machine learning models designed to use demographic and clinical predictors of IBD to predict which patients would fall into the “high healthcare utilization” category. Materials and Methods. A series of machine learning models were trained on a dataset extracted from a prospective natural history registry from a tertiary IBD center and associated healthcare charges. The models were trained to predict which patients are likely to have the highest healthcare utilization charges (top 15%). Results. A gradient-boosted trees classification model (accuracy 0.898, AUC 0.748) performed best out of the 12 evaluated modeling approaches. Conclusion. Classification models such as the ones evaluated in this work provide a reasonable basis for a clinical decision support system designed to predict which IBD patients are likely to become high expenditure.
Download
Paper Nr: 67
Title:

TAX-Corpus: Taxonomy based Annotations for Colonoscopy Evaluation
Authors:

Shorabuddin Syed, Adam J. Angel, Hafsa B. Syeda, Carole F. Jennings, Joseph VanScoy, Mahanazuddin Syed, Melody Greer, Sudeepa Bhattacharyya, Shaymaa Al-Shukri, Meredith Zozus, Fred Prior and Benjamin Tharian
Abstract: Colonoscopy plays a critical role in screening of colorectal carcinomas (CC). Unfortunately, the data related to this procedure are stored in disparate documents, colonoscopy, pathology, and radiology reports respectively. The lack of integrated standardized documentation is impeding accurate reporting of quality metrics and clinical and translational research. Natural language processing (NLP) has been used as an alternative to manual data abstraction. Performance of Machine Learning (ML) based NLP solutions is heavily dependent on the accuracy of annotated corpora. Availability of large volume annotated corpora is limited due to data privacy laws and the cost and effort required. In addition, the manual annotation process is error-prone, making the lack of quality annotated corpora the largest bottleneck in deploying ML solutions. The objective of this study is to identify clinical entities critical to colonoscopy quality, and build a high-quality annotated corpus using domain specific taxonomies following standardized annotation guidelines. The annotated corpus can be used to train ML models for a variety of downstream tasks.
Download
Paper Nr: 78
Title:

Context-aware Sleep Analysis with Intraday Steps and Heart Rate Time Series Data from Consumer Activity Trackers
Authors:

Zilu Liang, Huyen H. Nhung, Lauriane Bertrand and Nathan Cleyet-Marrel
Abstract: Wearable consumer activity trackers have become a popular tool for longitudinal monitoring of sleep quality. However, sleep data were routinely visualized in isolation from other contextual information. In this paper, we proposed a sleep analytics method to identify the associations between sleep quality and the contextual data that are readily measurable with a single Fitbit device. Different from prior studies that only focused on the daily aggregation of the contextual factors (e.g., total step counts), our method considers the intraday temporal patterns of these factors. Time-domain, frequency-domain, and nonlinear features were derived using the minute-by-minute intraday step and heart rate time series. The results showed that some of the identified contextual features such as the zero-crossing of steps and the absolute energy of heart rate could lead to actionable insights. While the nonlinear features—such as the average and longest diagonal line length derived through the recurrent quantitative analysis of the step time series—may not lead to insights that can be immediately acted on, they generated new hypotheses for further scientific studies. The results also showed that when dealing with data of consumer wearables, the individual-level analysis could generate more personally relevant insight than the cohort-level analysis.
Download
Paper Nr: 81
Title:

Vocabulary Modifications for Domain-adaptive Pretraining of Clinical Language Models
Authors:

Anastasios Lamproudis, Aron Henriksson and Hercules Dalianis
Abstract: Research has shown that using generic language models – specifically, BERT models – in specialized domains may be sub-optimal due to domain differences in language use and vocabulary. There are several techniques for developing domain-specific language models that leverage the use of existing generic language models, including continued and domain-adaptive pretraining with in-domain data. Here, we investigate a strategy based on using a domain-specific vocabulary, while leveraging a generic language model for initialization. The results demonstrate that domain-adaptive pretraining, in combination with a domain-specific vocabulary – as opposed to a general-domain vocabulary – yields improvements on two downstream clinical NLP tasks for Swedish. The results highlight the value of domain-adaptive pretraining when developing specialized language models and indicate that it is beneficial to adapt the vocabulary of the language model to the target domain prior to continued, domain-adaptive pretraining of a generic language model.
Download
Paper Nr: 85
Title:

The h-ANN Model: Comprehensive Colonoscopy Concept Compilation using Combined Contextual Embeddings
Authors:

Shorabuddin Syed, Adam J. Angel, Hafsa B. Syeda, Carole F. Jennings, Joseph VanScoy, Mahanazuddin Syed, Melody Greer, Sudeepa Bhattacharyya, Meredith Zozus, Benjamin Tharian and Fred Prior
Abstract: Colonoscopy is a screening and diagnostic procedure for detection of colorectal carcinomas with specific quality metrics that monitor and improve adenoma detection rates. These quality metrics are stored in disparate documents i.e., colonoscopy, pathology, and radiology reports. The lack of integrated standardized documentation is impeding colorectal cancer research. Clinical concept extraction using Natural Language Processing (NLP) and Machine Learning (ML) techniques is an alternative to manual data abstraction. Contextual word embedding models such as BERT (Bidirectional Encoder Representations from Transformers) and FLAIR have enhanced performance of NLP tasks. Combining multiple clinically-trained embeddings can improve word representations and boost the performance of the clinical NLP systems. The objective of this study is to extract comprehensive clinical concepts from the consolidated colonoscopy documents using concatenated clinical embeddings. We built high-quality annotated corpora for three report types. BERT and FLAIR embeddings were trained on unlabeled colonoscopy related documents. We built a hybrid Artificial Neural Network (h-ANN) to concatenate and fine-tune BERT and FLAIR embeddings. To extract concepts of interest from three report types, 3 models were initialized from the h-ANN and fine-tuned using the annotated corpora. The models achieved best F1-scores of 91.76%, 92.25%, and 88.55% for colonoscopy, pathology, and radiology reports respectively.
Download
Paper Nr: 87
Title:

A Review of Extended Reality Exercise Games for Elderly
Authors:

Yu Fu, Yan Hu, Veronica Sundstedt and Yvonne Forsell
Abstract: With the increasing of ageing all over the world, elderly health attracts more and more attention. This paper aims to study existing extended reality (XR) game applications for physical exercise through a literature review with 14 papers as an outcome. Based on these papers, we explored the contributions, opportunities, and challenges of exercise XR games for the elderly. The papers were analysed based on several perspectives, including publication information, design and implementation, game information, teamwork and social games, evaluation, and advantages and disadvantages. We found that the elderly were interested in and accepted the use of XR games. The positive effect of such games was common in the research results. Even if there were problems, such as simulator sickness, safety risks, device problems, and cost, there are still opportunities and space for research and development in the future. The overall positive attitudes toward XR exercise games for the elderly could be seen by both researchers, developers, and users. However, these game applications also presented some problems and future improvements are needed. The presented review is beneficial for researchers and developers to create or enhance future XR applications by learning from existing work.
Download
Paper Nr: 88
Title:

Predictive Modeling of Diabetes using EMR Data
Authors:

Hasan Zafari, Jie Li, Farhana Zulkernine, Leanne Kosowan and Alexander Singer
Abstract: As the prevalence of diabetes continues to increase globally, an efficient diabetes prediction model based on Electronic Medical Records (EMR) is critical to ensure the well-being of the patients and reduce the burden on the healthcare system. Prediction of diabetes in patients at an early stage and analysis of the risk factors can enable diabetes primary and secondary prevention. The objective of this study is to explore various classification models for identifying diabetes using EMR data. We extracted patient information, disease, health conditions, billing, and medication from EMR data. Six machine learning algorithms including three ensemble and three non-ensemble classifiers were used namely XGBoost, Random Forest, AdaBoost, Logistic Regression, Naive Bayes, and K-Nearest Neighbor (KNN). We experimented with both imbalanced data with the original class distribution and artificially balanced data for training the models. Our results indicate that the Random Forest model overall outperformed other models. When applied to the imbalanced data (112,837 instances), it results in the highest values in specificity (0.99) and F1-score (0.84), and when training with balanced data (35,858 instances) it achieves better values in sensitivity (1.00) and AUC (0.96). Analyzing feature importance, we identified a set of features that are more impactful in deciding the outcome including a number of comorbid conditions such as hypertension, dyslipidemia, osteoarthritis, CKD, and depression as well as a number of medication codes such as A10, D08, C10, and C09.
Download
Paper Nr: 90
Title:

Integrated Label Transfer for Oligodendrocyte Subpopulation Profiling in Parkinson’s Disease and Multiple System Atrophy
Authors:

Erin Teeple, Pooja Joshi, Rahul Pande, Yinyin Huang, Akshat Karambe, Martine Latta-Mahieu, S. P. Sardi, Angel Cedazo-Minguez, Katherine W. Klinger, Amilcar Flores-Morales, Stephen L. Madden, Deepak K. Rajpal and Dinesh Kumar
Abstract: Transfer of cell type labels as part of the comprehensive integration of multiple single nucleus RNA sequencing (snRNAseq) datasets offers a powerful tool for comparing cell populations and their activation states in normal versus disease conditions. Another potential use for these methods is annotation alignments between samples from different anatomic areas. This study describes and evaluates an integration analysis applied for profiling of oligodendrocyte lineage nuclei sequenced from human brain putamen region tissue samples for healthy Control (n = 3), Parkinson’s Disease (PD; n = 3) and Multiple System Atrophy (MSA; n = 3) subjects with label transfer to substantia nigra region tissue samples for healthy Control (n = 5) subjects. PD and MSA are both synucleinopathies, progressive neurodegenerative disorders characterized by nervous system aggregates of α-synuclein, a protein encoded by the SNCA gene. Histologic findings and genetic evidence suggest links between oligodendrocyte biology and synucleinopathy pathogenesis. In this work, we first identify disease-associated changes among transcriptionally distinct oligodendrocyte subpopulations in putamen. We then apply label transfer methods to generalize our findings from putamen to substantia nigra, a brain region characteristically impacted in PD and variably affected in MSA. Interestingly, our analysis predicts oligodendrocytes in substantia nigra include a significantly greater proportion of an oligodendrocyte subpopulation identified in putamen as most highly overexpressing SNCA in PD. Our results provide new insights into oligodendrocyte biology in PD and MSA and our workflow provides an example of label transfer methods applied for cross-dataset exploratory purpose.
Download
Paper Nr: 95
Title:

Simulation of the Evolution of a Virtual Patient’s Physiological Status in the Operating Room: Application to Computer-assisted Anaesthesia Training
Authors:

Hugo Boisaubert, Lucas Vincent, Corinne Lejus-Bourdeau and Christine Sinoquet
Abstract: Half a million surgeries are performed every day around the world, which places safety and quality at the heart of global health issues. In this context, we introduce a novel approach, SVP-OR (Simulation of Virtual Patient at the Operating Room), designed for digital training support. For this purpose, we must evolve the physiological parameters of a virtual patient submitted to the actions of a user (trainee), and of a virtual medical team. We formulate the problem as a case-based reasoning approach in which (i) we identify real patients whose anaesthetic profiles show a region similar to the recent history of the virtual patient and (ii) we predict the near future of the virtual patient (a multivariate time series) from the multivariate time series of the most similar real patients. The first contribution in this paper consists in the design of a contextualized multidimensional pattern recognition approach. Our second contribution is the development of a generic framework based on the concept of contextualized multidimensional pattern, to predict the evolution of the virtual patient. In a third contribution, we instantiate our framework, and we evaluate and compare the realism of two predictive strategies.
Download
Paper Nr: 98
Title:

Mmsd: A Multi-modal Dataset for Real-time, Continuous Stress Detection from Physiological Signals
Authors:

Mouna Benchekroun, Dan Istrate, Vincent Zalc and Dominique Lenne
Abstract: Although chronic stress is proven to be very harmful to physical and mental well being, its diagnosis is punctual and nontrivial, which calls for reliable, continuous and automated stress monitoring systems that do not yet exist. Wireless biosensors offer opportunities to remotely detect and monitor mental stress levels, enabling improved diagnosis and early treatment. There are different algorithms and methods for wearable stress detection, however, only a few standard and publicly available datasets exist today. In this paper, we introduce a multi-modal high-quality stress detection dataset with details of the experimental protocol. The dataset includes physiological, behavioural and motion data from 74 subjects during a lab study. Different modalities such as electrocardiograms (ECG), photoplethysmograms (PPG), electrodermal activity (EDA), electromyograms (EMG) as well as three axis gyroscope and accelerometer data were recorded. In addition, protocol validation was achieved using both subject’s self-reports and cortisol levels which is considered as gold standard for stress detection.
Download
Paper Nr: 99
Title:

Short-term Glucose Prediction based on Oral Glucose Tolerance Test Values
Authors:

Elias Dritsas, Sotiris Alexiou, Ioannis Konstantoulas and Konstantinos Moustakas
Abstract: Abnormal glucose metabolism increases the risk for cardiovascular disease and mortality. A key motivation for investigating this topic is Diabetes prevalence, which is the most common example of metabolic disorder that concern humans all over the world. The oral glucose tolerance test (OGTT) constitutes a traditional medical screening tool for all types of diabetes such as prediabetes, gestational, type 2 diabetes, insulin resistance or discrimination of Impaired Glucose Tolerance (IGT) from Natural Glucose Tolerance (NGT) individuals. Another motivation for this study is that a plethora of studies has shown the effectiveness of machine learning in glycemic control and improvement of diabetic’s management. This research study aims to evaluate the adequacy of machine learning on the short-term prediction of glucose levels. The main contribution of this analysis is a Random Forest regression tree model which, has been trained considering various risk factors and glucose samples obtained by a 2-hour OGTT, after a fast and then after an oral intake of glucose, at intervals of 30 minutes. The research outcomes verify the efficacy of Random Forest (RF).
Download
Paper Nr: 100
Title:

Vision-based Approach for Autism Diagnosis using Transfer Learning and Eye-tracking
Authors:

Mahmoud Elbattah, Jean-Luc Guérin, Romuald Carette, Federica Cilia and Gilles Dequen
Abstract: The potentials of Transfer Learning (TL) have been well-researched in areas such as Computer Vision and Natural Language Processing. This study aims to explore a novel application of TL to detect Autism Spectrum Disorder. We seek to develop an approach that combines TL and eye-tracking, which is commonly used for analyzing autistic features. The key idea is to transform eye-tracking scanpaths into a visual representation, which could facilitate using pretrained vision models. Our experiments implemented a set of widely used models including VGG-16, ResNet, and DenseNet. Our results showed that the TL approach could realize a promising accuracy of classification (ROC-AUC up to 0.78). The proposed approach is not claimed to provide superior performance compared to earlier work. However, the study is primarily thought to convey an interesting aspect regarding the use of (synthetic) visual representations of eye-tracking output as a means to transfer representations from models pretrained on large-scale datasets such as ImageNet.
Download
Paper Nr: 101
Title:

Ensuring Socio-technical Interoperability in Digital Health Innovation Processes: An Evaluation Approach
Authors:

Tim Scheplitz
Abstract: Integrating Digital Health Innovations (DHI) into healthcare practice remains a challenging task for innovators. They continuously seek for actionable ways to fulfil the complex web of requirements set by the target environment. A socio-technical understanding of interoperability offers structurization to this complexity and becomes a key property that innovators want to ensure during the innovation process. Nevertheless, scientific guidance remains abstract rather than applicable. This research paper builds on this point and follows the question how innovators can evaluate their DHI process holistically and tangibly to promote the later integration into complex healthcare systems. It therefore presents an evaluation approach based on the Refined eHealth European Interoperability Framework (ReEIF) and results of a qualitative content analysis. Here, detailed descriptions of the six ReEIF levels and 181 potential parameters for a self-assessment tool have been derived from prior literature. These findings stimulate future research on interdependencies within identified aspects of socio-technical interoperability and promote applicable tools for digital health innovators.
Download
Short Papers
Paper Nr: 2
Title:

Telehealth: A Viable Option for Optimizing Health System Performance during COVID-19: Call to Action for Future Pandemics
Authors:

Layal Mohtar and Nabil G. Badr
Abstract: Delve into the 21st century to welcome telehealth! It has taken so long coming, only to be accelerated by the COVID -19 pandemic. With the advent of telehealth solutions, healthcare systems are on the edge of the biggest gush in activity in over a century. In this paper, we look for evidence in the literature that treats the disruption introduced by Telehealth diffusion and the resulting, long awaited, contribution to optimizing health system performance. In our paper, we attempt to use the scoping review to detect evidence to answer this question. We performed a search up to April of 2021. Data were extracted on general study characteristics, clinical domain, technology, setting, category of outcome, and results. We then concluded with a synthesis of the information and call to action. We then coded the findings through the lens of the quadruple aim, provided reflections from the scoping review to inform how telehealth can be a dynamic element of system resilience. Though faced with unintended consequences, telehealth promises to be a viable alternative to in-person care, optimizing health system performance especially in times of constrained resources during a pandemic.
Download
Paper Nr: 3
Title:

Detection of Coughing and Respiratory Sensing in Conversational Speech
Authors:

Venkata S. Nallanthighal, Aki Harm, Ronald Rietman and Helmer Strik
Abstract: Coughing and shortness of breath are typical symptoms in people suffering from COPD, asthma, and COVID-19 conditions. Separate studies have shown that coughing and respiratory health parameters, respectively, can be sensed from a conversational speech recording using deep learning techniques. This paper looks into joint sensing of coughing events and the breathing pattern during natural speech. We introduce an algorithm and demonstrate its performance in realistic recordings. We observed sensitivity of 92.4% and 91.6% for cough detection and breath event detection, respectively. Clinical Relevance: Joint sensing of coughing events and respiratory parameters gives a more holistic picture of the respiratory health of a patient which can be very useful for future telehealth services.
Download
Paper Nr: 4
Title:

Designing Neurogames to Support Patients under Psychotherapy Treatment: Opportunities and Challenges of the MUSE Headband
Authors:

Michael Pantförder and Andre Hellwig
Abstract: In childhood and adolescence, mental development processes are crucial for a person’s long-term, cognitive health. Many young people have at least one characteristic that leads to psychological impairment and must be accompanied by therapy. Therapy success requires the constant execution of therapeutic exercises during and after therapy sessions. However, keeping the motivation of the patients upright for continuous cooperation is a key challenge, since the exercises are perceived as laborious. A digital, playful training application offers the potential to support the therapy of children and adolescents. Measuring brain activity plays an important role as it shows how good patients can push away negative thoughts affecting their mental disorder. For this purpose the fundamentals of serious games, neurofeedback, brain computer interfaces (BCIs) and electroencephalography (EEG) as well as different therapy-accompanying measures were examined. Based on the findings and a focus group with psychotherapists (N=3), a serious game was designed as a motivational concentration and attention training to support psychotherapy. During the game the easy-to-use MUSE headband measures concentration and integrates neurofeedback as a game mechanic. User tests with children (N=21) were performed to evaluate the developed prototype and gather further information on usability, technology acceptance and playfulness of the neurogame.
Download
Paper Nr: 10
Title:

Extracting Alarm Events from the MIMIC-III Clinical Database
Authors:

Jonas Chromik, Bjarne Pfitzner, Nina Ihde, Marius Michaelis, Denise Schmidt, Sophie I. Klopfenstein, Akira-Sebastian Poncette, Felix Balzer and Bert Arnrich
Abstract: Lack of readily available data on ICU alarm events constitutes a major obstacle to alarm fatigue research. There are ICU databases available that aim to give a holistic picture of everything happening at the respective ICU. However, these databases do not contain data on alarm events. We utilise the vital parameters and alarm thresholds recorded in the MIMIC-III database in order to artificially extract alarm events from this database. Prior to that, we uncover, investigate, and mitigate inconsistencies we found in the data. The results of this work are an approach and an algorithm for cleaning the alarm data available in MIMIC-III and extract concrete alarm events from them. The data set generated by this algorithm is investigated in this work and can be used for further research into the problem of alarm fatigue.
Download
Paper Nr: 13
Title:

Harena Semantics: A Framework to Support Semantic Annotation in Citizen Science Systems
Authors:

Fagner L. Pantoja, Marco C. Filho and André Santanchè
Abstract: We propose a new approach to support human agents to annotate semantic concepts in free-text sentences in the biomedical domain. Using our markdown-derived language called Versum, authors can easily annotate relevant terms while producing content for Citizen Science systems. Besides, an embedded Automatic Annotation Mechanism suggests semantic concepts for the author. It implements a Named Entity Recognition task using a hybrid approach: (1) a Transformer-based Deep Neural Network and (2) an Ontology-based method. We conducted a case study running over content produced in the Harena e-learning system, which intends to teach Clinical Reasoning to students using Clinical Cases. Results of this pilot evaluation suggest the potential of Harena Semantics to engage volunteers in the production of semantic, agent-centered resources on crowdsourcing systems.
Download
Paper Nr: 14
Title:

Impact of Machine Learning Assistance on the Quality of Life Prediction for Breast Cancer Patients
Authors:

Mikko Nuutinen, Sonja Korhonen, Anna-Maria Hiltunen, Ira Haavisto, Paula Poikonen-Saksela, Johanna Mattson, Haridimos Kondylakis, Ketti Mazzocco, Ruth Pat-Horenczyk, Berta Sousa and Riikka-Leena Leskelä
Abstract: Proper and well-timed interventions may improve breast cancer patient adaptation, resilience and quality of life (QoL) during treatment process and time after disease. The challenge is to identify those patients who would benefit most from a particular intervention. The aim of this study was to measure whether the machine learning prediction incorporated in the clinical decision support system (CDSS) improves clinicians’ performance to predict patients’ QoL during treatment process. We conducted an experimental setup in which six clinicians used CDSS and predicted QoL for 60 breast cancer patients. Each patient was evaluated both with and without the aid of machine learning prediction. The clinicians were also open-ended interviewed to investigate the usage and perceived benefits of CDSS with the machine learning prediction aid. Clinicians’ performance to evaluate the patients’ QoL was higher with the aid of machine learning predictions than without the aid. AUROC of clinicians was .777 (95% CI .691 − .857) with the aid and .755 (95% CI .664 − .840) without the aid. When the machine learning model’s prediction was correct, the average accuracy (ACC) of the clinicians was .788 (95% CI .739 − .838) with the aid and .717 (95% CI .636 − .798) without the aid.
Download
Paper Nr: 15
Title:

Salting as a Countermeasure against Attacks on Privacy Preserving Record Linkage Techniques
Authors:

Yanling Chen, Rainer Schnell, Frederik Armknecht and Youzhe Heng
Abstract: Privacy-preserving record linkage (PPRL) is the research area dedicated to linking records from multiple databases for the same patient without revealing identifying information during the linkage. A popular PPRL approach is based on Bloom filters (BF). Recent research has shown that BF based PPRL could be vulnerable to cryptanalysis attacks. Among several hardening techniques, salting was considered to be one of the most suitable defences. A thorough evaluation of the amount of protection provided by salting is lacking from the literature. In this paper, we empirically evaluate the effect of salting on privacy by demonstrating the resilience of salted BF to the two most advanced attack methods: pattern mining and graph-matching. Experimental results show that salting could improve resilience against both attacks, although more minor against graph-matching attacks than pattern mining attacks.
Download
Paper Nr: 16
Title:

Virtual Reality Ontology Object Manipulation (VROOM)
Authors:

Margarita Vinnikov, Daniel Vergilis, Uras Oran and James Geller
Abstract: Biomedical ontologies are considered important repositories of knowledge about the medical domain and related fields. They are best thought of as node-link networks, where each node represents one single (medical) concept and links express binary relationships between pairs of concepts. The most important relationship (“IS-A”) is used to form a generalization hierarchy among the concepts. Visualizing and manipulating such a network when it includes more than a few hundred nodes can be challenging. This paper presents a new system called VROOM (Virtual Reality Ontology Object Manipulation) that supports browsing and interaction with a biomedical ontology in a virtual 3-D space, enabling more natural and realistic interaction with and navigation through the ontology network.
Download
Paper Nr: 20
Title:

Detection and Remediation of Malicious Actors for Studies Involving Remote Data Collection
Authors:

Bethany K. Bracken, John Wolcott, Isaac Potoczny-Jones, Brittany A. Mosser, Isabell R. Griffith-Fillipo and Patricia A. Arean
Abstract: Although most human subjects research requires data collection by contacting local participants who visit a research site, some studies require increasingly large troves of data collected continuously during their typical daily lives using sensors (e.g., fitness trackers) and ecological momentary assessments. Long-term, continuous collection is becoming more feasible as smartphones become ubiquitous. To enable remote collection of these rich data sets while ensuring privacy, we built a system to allow secure and fully human-out-of-the-loop participant recruitment, screening, onboarding, data collection on smartphones, data transmission to the cloud, data security in the cloud, and data access by analysis and modeling teams. Study participants were paid for completion of daily ecological momentary assessments in keeping with standards of research equipoise, fairness, and retention strategies. However, our study attracted “malicious actors” who were pretending to be study participants, but were not, in order to receive payment. This opinion piece outlines how we initially detected malicious actors, and the steps we took in order to prevent future malicious actors from enrolling in the study. This resulted in several lessons learned that we think will be valuable for future studies that recruit, enroll, and maintain study participants remotely.
Download
Paper Nr: 24
Title:

Internet of Health Things for Quality of Life: Open Challenges based on a Systematic Literature Mapping
Authors:

Pedro M. Oliveira, Rossana C. Andrade, Pedro S. Neto and Breno S. Oliveira
Abstract: Internet of Health Things (IoHT) related papers has produced valuable knowledge concerning applications such as monitoring vital signs and predicting diseases. However, this knowledge is dispersed in the literature and, to the best of our knowledge, we could not find a recent study summarizing it. Thus, this work presents a systematic mapping conducted to organize the challenges regarding IoHT applied to QoL. As a result, we highlight a growing interest in developing health monitoring tools, but without many real-world validations. The most mentioned challenges were well-known IoT challenges, security and privacy, data science, and networks. Moreover, despite many studies discussing proposals for improving QoL, few papers sought to measure this gain, and none addressed the semantic organization of QoL data obtained from smart objects. Finally, it is expected for future a strengthening regarding elderly healthcare solutions, data science usage for personalized systems; smart models to predict health problems; and QoL continuous monitoring.
Download
Paper Nr: 25
Title:

e-Health Services to Support the Perinatal Decision-making Process: An Analysis of Digital Solutions to Create Birth Plans
Authors:

Carla V. Leite and Ana M. Almeida
Abstract: This research aims to provide an overview of the existent digital solutions for birth plans’ creation, intending to contribute for the advance of e-health services focused on the perinatal decision-making process. Primary data was found through a web search procedure. Better ranked options complying with the following criteria were included: (a) available online and for free; (b) pregnant people as the target audience; (c) labor and/or birth plan creation features; (d) in English. Four online services were found, and a two part study was conducted: a) a non-exhaustive benchmarking-like analysis of webpages where the digital solutions to create birth plans were provided, according to six dimensions; b) followed by a content analysis of the digital solutions, resulting in 13 categories emerging, that were scored according to their occurrence and completeness. “Consent and Information” category had the lowest score, what is considered critical for the full purpose of a birth plan creation; while, “Freedom”, “Ambience and Equipment”, “People”, “Type of birth” and “Pain management” categories achieved the highest scores. Two solutions were considered particularly incomplete. Results show three solutions based on checklists, and one on visual icons. All solutions were based on a delivery approach, not including interactive or audiovisual components.
Download
Paper Nr: 26
Title:

A Review of the Main Factors, Computational Methods, and Databases Used in Depression Studies
Authors:

Ariane C. B. da Silva, Renata C. Santana, Thiago H. N. de Lima, Maycoln M. Teodoro, Mark A. Song, Luis E. Zárate and Cristiane N. Nobre
Abstract: Depression is a mental health disorder that affects millions of people worldwide. The disorder results from a complex interaction of biological, psychological, and social factors, leading to difficulty in both prognosis and diagnosis. In this work, we performed a review on studies about depression, to identify the main computational techniques used to support the prediction (prognosis and diagnosis) of depression, and the main attributes that might influence the development of the disorder. Our results indicate that, in the last ten years, Logistic Regression, Machine Learning techniques such as Support Vector Machines and Neural Networks, and other supervised learning algorithms, have been employed more frequently for studies predicting depression and selecting features related to it. Attributes like insomnia, gender, marital state, and use of tobacco, for example, were related to the development of depression. The review indicated growing effectiveness in using machine learning methods for analyzing data related to depression.
Download
Paper Nr: 27
Title:

How Health Information Spreads in Twitter: The Whos and Whats of Philippine TB-data
Authors:

Erika Y. Chan, Myles C. Chan, Shyrene S. Ching, Stanley L. Sie, Angelyn R. Lao, Jan C. Bernadas and Charibeth K. Cheng
Abstract: Twitter is a popular platform for disseminating health information. Unfortunately, there is no clear way to monitor how information reaches the intended audiences. This research examined how health information spreads on Twitter and identified factors that affect the spreading within the Philippines. We created a process whose goal is to generate results that experts can deeply analyze to reveal insights into information spread. The process consists of crawling Twitter data, transforming the data and applying sentiment identification and topic modeling, and performing Social Network Analysis (SNA). The SNA graphs allow for the study of the interactions between Twitter users and tweets while giving insights on influential users and topics discussed across clusters. The study explored and utilized tuberculosis-related tweets. Though the algorithms were meant to process tweets written in Filipino, the process is mostly language-agnostic and can be applied to Twitter data. The results also help in identifying strategies that can improve health information spread on Twitter in the Philippines.
Download
Paper Nr: 30
Title:

Simulating the Doctor’s Behaviour: A Preliminary Study on the Identification of Atrial Fibrillation through Combined Analysis of Heart Rate and Beat Morphology
Authors:

Gennaro Laudato, Giovanni Rosa, Giovanni Capobianco, Angela R. Colavita, Arianna D. Forno, Fabio Divino, Claudio Lupi, Remo Pareschi, Stefano Ricciardi, Luca Romagnoli, Simone Scalabrino, Cecilia Tomassini and Rocco Oliveto
Abstract: Atrial fibrillation (AF) is a medical disorder that affects the atria of the heart. AF has emerged as a worldwide cardiovascular epidemic affecting more than 33 million people around the world. Several automated approaches based on the analysis of the ECG have been proposed to facilitate the manual identification of AF episodes. Especially, such approaches analyze the heartbeat morphology (absence of P-wave) or the heart rate (presence of arrhythmia). In this article, we present AMELIA (AutoMatic dEtection of atriaL fIbrillation for heAlthcare), an approach that simulates the doctor’s behavior by considering both the sources of information in a combined way. AMELIA is basically composed of two components; one integrating a LSTM (Long Short-Term Memory) Recurrent Neural Network (RNN) and the second integrating a rhythm analyzer. When the RNN reveals a heartbeat with abnormal morphology, the rhythm analyzer is activated to verify whether or not there is a simultaneous arrhythmia. AMELIA has been experimented by using well-known datasets, namely Physionet-AF and NSR-DB. The achieved results provide evidence of the potential benefits of the approach, especially regarding sensitivity. AMELIA has an incredibly high potential to be used in applications of continuous monitoring, where the detection of AF episodes is a fundamental and crucial activity.
Download
Paper Nr: 32
Title:

Risk-based Comprehensive Usability Evaluation of Software as a Medical Device
Authors:

Noemi Stuppia, Federico Sternini, Federica Miola, Giorgia Picci, Claudia Boarini, Federico Cabitza and Alice Ravizza
Abstract: Introduction: Usability evaluation is a core aspect in risk assessment of medical devices, as it aims to ensure the device interface safety, avoiding that usability problems at interface level are not related to harm. Methods: Our research group applied our risk-based approach, international reference standards and guidelines to the usability evaluation of a large family of SaMD. The methodology used for the evaluation is an elaboration of regulatory prescriptions and is composed of a combination of quantitative and qualitative methods. In particular, the usability evaluation is structured in a two-stage evaluation composed by formative and summative evaluation. The formative stage is propaedeutic for the planning of the summative evaluation. The final assessment included the analysis of quantitative data collected through three questionnaires and a user test. Results and discussion: Risk-based task analysis led to the identification of the most common use error emerged during the user test performance. The three questionnaires led to different results: Heuristic analysis allowed the identification of violations to the heuristic principles as perceived by the users and their severity; SUS questionnaire provided an indicator of general device usability; the interview identified the usability problems of each device with respect to their functionalities. Conclusions: The study allowed the extensive assessment of the devices, the identification of usability issues, and the classification in terms of criticality of each issue. In conclusion the study led to different proposals to solve the issues and design changes.
Download
Paper Nr: 34
Title:

Using Merged Cancer Registry Data for Survival Analysis in Patients Treated with Integrative Oncology: Conceptual Framework and First Results of a Feasibility Study
Authors:

Thomas Ostermann, Sebastian Appelbaum, Stephan Baumgartner, Lukas Rist and Daniel Krüerke
Abstract: Survival analysis is the basis for research into all types of treatments aimed at prolonging the overall survival of a cancer entity. Before we use data from a cancer registry at the Clinic Arlesheim (CRCA) for more sophisticated survival analysis in relation to integrative oncology treatments, we wanted to learn more about the possible differences between the clientele in this database and the public. In a first step we compared survival rates for breast cancer and pancreatic cancer analyzed from CRCA-data with the cor-responding survival rate (all stages) available at the Robert-Koch-Institute. Furthermore, we differentiated the survival rates from CRCA-patients with respect to the fraction of the survival time in the care of the clinic Arlesheim. While the survival rates of CRCA-patients with breast cancer or with pancreatic cancer show similar survival rates compared to corresponding data from the Robert-Koch-Institute, the sensitivity analysis suggests that the longer the fraction of the survival time in the care of the clinic Arlesheim the higher the expected survival rates. In conclusion, the analysis and comparison of the survival rates of a clinical population of a cancer registry, such as CRCA, may lead to a better identification of responders and non-responders and thus in the long run may help to optimise integrative and patient cantered treatment strategies.
Download
Paper Nr: 35
Title:

Implementation and Feasibility Analysis of a Javascript-based Gambling Tool Device for Online Decision Making Task under Risk in Psychological and Health Services Research
Authors:

Sherine Franckenstein, Sebastian Appelbaum and Thomas Ostermann
Abstract: Decision making is one of the most complex tasks in human behavior. In the past, researchers have tried to understand how humans make decisions by designing neuropsychological tests to assess reward related decision making by evaluating the preference for smaller but immediate rewards over larger but delayed rewards or by evaluating the tolerance of risk in favor of a desired reward. The latter are also known as gambling tasks. Today, information technology offers a variety of possibilities to investigate behaviour under risk. After a short introduction on gambling tasks and in particular the game of dice task, this article describes the development and implementation of a JavaScript-based gambling tool for online surveys based on a game of dice task. In a pilot feasibility study with 170 medical students, participants were randomly assigned to a “REAL condition”, based on the probabilities of the chosen bet and a “FAKE condition” where participants lose all the time independently of the chosen bet. We were able to show that the software was well accepted with only 14.7% of drop outs. Moreover, we also found a difference between the FAKE and the REAL group: Participants in the FAKE condition in the mean steadily increased their stake while then control group quite early tended to run a safer strategy. This is also obvious when the overall stake mean is compared: While in the REAL condition the mean stake is 310.89 ± 222.98 €, the FAKE condition has an overall mean of 390.38 ± 296.50 €. In conclusion, this article clearly indicates how a JavaScript based gambling tool can be used for psychological online research.
Download
Paper Nr: 37
Title:

Predictive Tools to Evaluate Cardiovascular Events in Chronic Heart Failure Patients
Authors:

Maria Carmela Groccia, Rosita Guido, Domenico Conforti and Angela Sciacqua
Abstract: In this paper, a Knowledge Discovery task has been implemented with the aim of developing models for predicting cardiovascular worsening events in Chronic Heart Failure (CHF) patients. A set of patients suffering from CHF were enrolled and carefully evaluated through a five-year follow-up. Several predictive models were developed on the collected data and then compared. Among these, the decision tree based predictive model has been analysed by clinical experts. The decision tree is among all the trained and tested models the most simple and interpretable mainly by clinicians because it discovers if-then rules. The extracted rules are compliant with previous clinical studies. Nevertheless, the decision tree achieved lower performance compared to the other predictive models, which conversely to the decision tree are not “clinician friendly” because they do not provide an explanation of the classification decisions.
Download
Paper Nr: 40
Title:

Dashboard for Machine Learning Models in Health Care
Authors:

Wejdan Bagais and Janusz Wojtusiak
Abstract: To trust and use machine learning (ML) models in health settings, decision-makers need to understand the model's performance. Yet, there has been little agreement on what information should be visualized to present models' evaluations. This work presents an approach to construct a dashboard used to visualize supervised ML models for health care applications. The dashboard shows the models' statistical evaluations, feature importance, and sensitivity analysis.
Download
Paper Nr: 42
Title:

From Wearable Device to OpenEMR: 5G Edge Centered Telemedicine and Decision Support System
Authors:

Ying Wang, Patricia Tran and Janusz Wojtusiak
Abstract: The Internet of Things (IoT) is developing rapidly, with applications across various fields and industries. In healthcare, wearable devices and the Internet of Medical Things (IoMT) have tremendous potential for improvements in the quality of telemedicine and producing medical insights and discoveries. Massive Machine Type of Communication (mMTC) in 5G further reduces latency and enhances connectivity in supporting wearables and IoMT, which provides a promising infrastructure for telemedicine. Although cloud computing reduced the computation and storage load on wearable devices significantly, the massive amounts of data produced by wearable devices and IoMT introduce challenges for latency and storage in the cloud. Additionally, applications will need to navigate the regulation and compliance laws related to handling sensitive and private health data, adding complexity to the accessibility and distribution of such innovations. This study first examined the current frameworks for wearable devices in 5G telemedicine implementation and discussed existing challenges. We then proposed a multi-layer 5G mobile edge computing (MEC) centered telemedicine design that dynamically integrates wearable devices with OpenEMR electronic health records system. The multi-layer design includes the IoT layer, MEC layer, Network layer, and Application layer. Near-real-time artificial intelligence (AI) components and electronic health record (EHR) instances are automatically deployed to and removed from the MEC layer to keep cloud computing capabilities closest to the infrastructure edge when a user is associating and disassociating with a 5G bases station, respectively. Lastly, we demonstrate a proof of concept by designing and implementing a system for detecting atrial fibrillation (Afib) over the design we proposed. Afib detection has the character of predictable trending, random occurrence of adverse events, and urgent care needed when happening. These characters requires a low latency, large range coverage and high throughput infrastructure. The proposed approach provides a distributed solution addressing the requirements for Afib detection. This approach can be used for other applications in telemedicine beyond Afib detection.
Download
Paper Nr: 44
Title:

Forecasting Emergency Department Crowding using Data Science Techniques
Authors:

José Manuel Domenech Cabrera and Javier Lorenzo-Navarro
Abstract: The provision of insufficient resources during periods of high demand can lead to overcrowding in emergency departments. This issue has been extensively addressed through time series forecasting and regression problems. Despite the fact the increasing number of studies, accurate forecasting of demand remains a challenge. Thus, the purpose of this study was to develop a tool to predict the future evolution of emergency department occupancy in order to anticipate overcrowding episodes, avoid their negative effects on health and improve efficiency. This article presents a novel approach under the premise that the ability of the system to drain patients is the most determining factor in overcrowding episodes as opposed to previous approaches focused on patient demand. The forecasts model were based on the hourly number of patients occupying the general Emergency Department of Insular University Hospital of Gran Canaria Island, mainly given data of the flow of patients through the emergency department as well as performance indicators from other areas of the hospital extracted from the information system.
Download
Paper Nr: 45
Title:

Predicting Depression in Children and Adolescents using the SHAP Approach
Authors:

Marcelo Balbino, Renata Santana, Maycoln Teodoro, Mark Song, Luis Zárate and Cristiane Nobre
Abstract: Depression is a disease with severe consequences that affects millions of people, with the onset of the first symptoms being common in youth. It is essential to identify and treat individuals with depression as early as possible to prevent the losses caused by the disorder throughout life. However, the diagnostic criteria of depressive disorders for children/adolescents or adults is not differentiated, even though authors claim that the particularities of childhood must be considered. This may be why childhood depression is being underdiagnosed. Therefore, this work aims to discover the most significant features in diagnosing depression in children and adolescents through Machine Learning methods and the SHAP approach. Models with Machine Learning algorithms were developed, and the model with SVM presented the best results. The application of SHAP proved to be fundamental to deepen the understanding of this model. The experiments indicated that feelings of isolation, sadness, excessive worry, complaints about one’s appearance, resistance to academic tasks, and the mother’s schooling are the most significant features in predicting depression in children and adolescents. Such results can help to understand depression in these individuals and thus lead to appropriate treatment.
Download
Paper Nr: 50
Title:

A Health IT-Empowered Integrated Platform for Secure Vaccine Data Management and Intelligent Visual Analytics and Reporting
Authors:

Jay Patel, Bari Dzomba, Hoa Vo, Susan Von Nessen-Scanlin, Laura A. Siminoff and Huanmei Wu
Abstract: Health IT (HIT) and big data analysis have been applied to a community-oriented COVID-19 vaccination program (RapidVax). The HIT platform enables security data collection, enforces data quality and rule validations, preserves privacy through strict data access control with HIPAA compliance and secure VPN, customizes interactive user interfaces, empowers outcome visualization, and generates intelligent reporting. The RapidVax program has adopted the HIT platform for ninety-five vaccination events in thirty geographically separated communities. Our study demonstrated the significance of health IT tools, and automated program generated in this study to help manage a public health problem such as the COVID-19 pandemic. The health IT tools developed in this study provided an essential piece of critical infrastructure which supported our clinicians to run the vaccination task efficiently.
Download
Paper Nr: 51
Title:

Have Your Cake and Log it Too: A Pilot Study Leveraging IMU Sensors for Real-time Food Journaling Notifications
Authors:

Arpita Kappattanavar, Marten Kremser and Bert Arnrich
Abstract: To monitor diet, nutritionists employ food journaling approaches, which rely on the subject’s memory. Accordingly, a real-time reminder during eating can help subjects adhere to a journaling routine more strictly. Although previous works used sensors to detect eating activities, no study accounted for the time impact of delivering notifications. Our study presents an approach to notify subjects for food journaling within three to six minutes from eating. We achieved this by collecting wrist motion data using an inertial measurement unit. Twenty-two features were extracted from the collected data. Those were used as input to a random forest model to classify an eating activity. To train and test the model, we collected data from four subjects in a semi-controlled environment and daily life. The f1-score for testing data was between 0.74 to 0.78 for four subjects, but they still received notifications for all meals. Additionally, we tested this approach with data collected for one and a half days from a new subject. We observed notifications for four out of five meals. The robust detection criterion reduced the false notifications. Our pilot study results suggest that considering the delivery time of notification can lead to better food journaling
Download
Paper Nr: 61
Title:

mHealth Use in Healthcare Facilities: Raising Awareness in Data Protection, Privacy and Safety
Authors:

Lilian M. Ader, Bróna MacEntee, Kristina Rutkauskaite, Nutsa Chichilidze, Dylan Kearney, Sean A. Lynch, Katie Crowley and Ita Richardson
Abstract: During the COVID-19 pandemic, many patients and healthcare professionals embraced the possibility of using available mobile devices and applications, exploring the opportunities to reduce the burden on strained services. However, despite strict surveillance under the European GDPR or Medical Device (MD) regulations, users are considered to be primarily responsible for verifying that their application of choice is approved and certified. We searched academic and grey literature and discuss some of the challenges related to the use of personal devices and mobile applications for health and medical purposes. Our position is that policies and technologies should be more considerate of users’ behaviour, which includes use of non-medical software for medical purposes, and situations where users seem to choose usability over safety.
Download
Paper Nr: 64
Title:

A Trusted Data Sharing Environment based on FAIR Principles and Distributed Process Execution
Authors:

Marcel Klötgen, Florian Lauf, Sebastian Stäubert, Sven Meister and Danny Ammon
Abstract: Provision and usage of distributed secondary-use data for medical research requires the implementation of a distributed data use & access process and several sub-processes. The SMITH Service Platform (SSP) manages process-based interactions with several Data Integration Centers (DIC), each being responsible for the management and provision of suitable data sets in the context of a data use project. A trusted data sharing environment is specified based on the implementation of Trusted Connectors as specified by the International Data Spaces (IDS) Reference Architecture Model. Thus, a distributed data delivery sub-process enforces data sovereignty aspects between all involved parties. In the future, a solidification of the concepts and a further implementation of the trusted data sharing environment should be addressed.
Download
Paper Nr: 65
Title:

Intra-individual Stability and Assessment of the Affective State in a Virtual Laboratory Environment: A Feasibility Study
Authors:

Nils M. Vahle, Sebastian Unger and Martin J. Tomasik
Abstract: While virtual reality (VR) emerges in a variety of research contexts, the effects on behavior and performance caused by VR-based embodiment still lack sufficient evidence of changes in affective state. With this feasibility study, we compared the affective states in both younger and older adults, measured after conventional computer-based tests in real life (RL) and after tests in VR. These assessment tests are spread over five time points, two in RL and three in VR, and the differences between the VR and the RL environment are investigated against the backdrop of two theoretical models of cognitive psychology. Results showed no change in affective state in either age group, switching from a RL to a VR environment. In addition, the elderly did not assess their affective state significantly different than that of the younger control group. In conclusion, lifelike VR environments for cognitive testing and other assessment or training purposes do not seem to lead to any systematic influence of affective state compared to RL computer-based assessments, making VR an alternative to conventional methods, for instance for cognitive treatments or preventions. Although the results can only be partially generalized due to a small sample size, they show technical stability and suitability for future use of similar applications.
Download
Paper Nr: 69
Title:

A Process Cube based Approach of Process Mining in Analysing Frailty Progression Exploiting Electronic Frailty Index
Authors:

N. F. Farid, Marc de Kamps and Owen A. Johnson
Abstract: Process mining is a data analytics technique that is used in healthcare to develop insights into care processes, care pathways and disease progression using event data extracted from Health Information Systems. The most widely used application is process discovery where models of healthcare processes are automatically inferred and visualized. These have been applied to frailty, a common geriatric condition in elderly people typically described in terms of progression through a number of stages. In this paper we use the Electronic Frailty Index which is calculated using 36 indicators of frailty deficits. We use process mining to analyse frailty progression using data from the SystmOne GP system used in UK primary care. We propose an approach for analysing frailty progression using a process cube analysis through slicing and dicing sets of attributes related to clinical frailty events. Different combinations of process cube dimensions allow us to model and analyse a comprehensible frailty progression. We illustrate the method through a case study investigating the association between frailty stages and three common issues; falls, hypertension and polypharmacy.
Download
Paper Nr: 70
Title:

Towards an Ethical Framework for the Design and Development of Inclusive Home-based Smart Technology for Older Adults and People with Disabilities
Authors:

Emma Murphy, Damian Gordon, Brian Keegan, Julie Doyle, Ioannis Stavrakakis and Dympna O’Sullivan
Abstract: Unique ethical, privacy and safety implications arise for people who are reliant on home-based smart technology due to health conditions or disabilities. In this position paper we highlight a need for a reflective, inclusive ethical framework that encompasses the life cycle of smart home technology design. We present key ethical considerations in the design, development and deployment of smart home-based technology for older adults and people with disabilities. Using ethical theories, human-centred design and personas we explore how some of these critical issues can be addressed. Finally, we propose a novel ethical framework for the development of inclusive home-based smart technology which combines these key considerations with existing models of design.
Download
Paper Nr: 71
Title:

Sovereignly Donating Medical Data as a Patient: A Technical Approach
Authors:

Florian Lauf, Hendrik Z. Felde, Marcel Klötgen, Robin Brandstädter and Robin Schönborn
Abstract: Data is the new asset of the 21st century, and many new business models are based on data. However, data is also needed in the medical research domain, such as in the procedure of applying new machine learning methods for gaining new medical findings. Furthermore, the hurdle arises that medical data comprises personal data, and thus, it requires particular care and protection. Hence, patients must consent to the data donation process for general medical research but without selecting specific research projects. We argue that patients must gain more influence in the data donation process to cover this lack of data sovereignty. Therefore, we developed a concept and implementation empowering patients to make sovereign decisions about donating their medical data to specific medical research projects. Our work comprises concepts of the Medical Informatics Initiative, International Data Spaces, and MY DATA Control Technologies with new specific elements combining these components. This approach of patient empowerment enables a new kind of data sovereignty in the medical research domain.
Download
Paper Nr: 73
Title:

Understanding Public Opinion on using Hydroxychloroquine for COVID-19 Treatment via Social Media
Authors:

Thuy T. Do, Du Nguyen, Anh Le, Anh Nguyen, Dong Nguyen, Nga Hoang, Uyen Le and Tuan Tran
Abstract: Hydroxychloroquine (HCQ) is used to prevent or treat malaria caused by mosquito bites. Recently, the drug has been suggested to treat COVID-19, but that has not been supported by scientific evidence. The information regarding the drug efficacy has flooded social networks, posting potential threats to the community by perverting their perceptions of the drug efficacy. This paper studies the reactions of social network users on the recommendation of using HCQ for COVID-19 treatment by analyzing the reaction patterns and sentiment of the tweets. We collected 164,016 tweets from February to December 2020 and used a text mining approach to identify social reaction patterns and opinion change over time. Our descriptive analysis identified an irregularity of the users’ reaction patterns associated tightly with the social and news feeds on the development of HCQ and COVID-19 treatment. The study linked the tweets and Google search frequencies to reveal the viewpoints of local communities on the use of HCQ for COVID-19 treatment across different states. Further, our tweet sentiment analysis reveals that public opinion changed significantly over time regarding the recommendation of using HCQ for COVID-19 treatment. The data showed that high support in the early dates but it significantly declined in October. Finally, using the manual classification of 4,850 tweets by humans as our benchmark, our sentiment analysis showed that the Google Cloud Natural Language algorithm outperformed the Valence Aware Dictionary and sEntiment Reasoner in classifying tweets, especially in the sarcastic tweet group.
Download
Paper Nr: 74
Title:

DeIDNER Model: A Neural Network Named Entity Recognition Model for Use in the De-identification of Clinical Notes
Authors:

Mahanazuddin Syed, Kevin Sexton, Melody Greer, Shorabuddin Syed, Joseph VanScoy, Farhan Kawsar, Erica Olson, Karan Patel, Jake Erwin, Sudeepa Bhattacharyya, Meredith Zozus and Fred Prior
Abstract: Clinical named entity recognition (NER) is an essential building block for many downstream natural language processing (NLP) applications such as information extraction and de-identification. Recently, deep learning (DL) methods that utilize word embeddings have become popular in clinical NLP tasks. However, there has been little work on evaluating and combining the word embeddings trained from different domains. The goal of this study is to improve the performance of NER in clinical discharge summaries by developing a DL model that combines different embeddings and investigate the combination of standard and contextual embeddings from the general and clinical domains. We developed: 1) A human-annotated high-quality internal corpus with discharge summaries and 2) A NER model with an input embedding layer that combines different embeddings: standard word embeddings, context-based word embeddings, a character-level word embedding using a convolutional neural network (CNN), and an external knowledge sources along with word features as one-hot vectors. Embedding was followed by bidirectional long short-term memory (Bi-LSTM) and conditional random field (CRF) layers. The proposed model reaches or overcomes state-of-the-art performance on two publicly available data sets and an F1 score of 94.31% on an internal corpus. After incorporating mixed-domain clinically pre-trained contextual embeddings, the F1 score further improved to 95.36% on the internal corpus. This study demonstrated an efficient way of combining different embeddings that will improve the recognition performance aiding the downstream de-identification of clinical notes.
Download
Paper Nr: 76
Title:

Feature Selection for Sentiment Classification of COVID-19 Tweets: H-TFIDF Featuring BERT
Authors:

Mehtab A. Syed, Elena Arsevska, Mathieu Roche and Maguelonne Teisseire
Abstract: In the first quarter of 2020, the World Health Organization (WHO) declared COVID-19 a public health emergency around the globe. Different users from all over the world shared their opinions about COVID-19 on social media platforms such as Twitter and Facebook. At the beginning of the pandemic, it became relevant to assess public opinions regarding COVID-19 using data available on social media. We used a recently proposed hierarchy-based measure for tweet analysis (H-TFIDF) for feature extraction over sentiment classification of tweets. We assessed how H-TFIDF and concatenation of H-TFIDF with bidirectional encoder representations from transformers (BH-TFIDF) perform over state-of-the-art bag-of-words (BOW) and term frequency-inverse document frequency (TF-IDF) features for sentiment classification of COVID-19 tweets. A uniform experimental setup of the training-test (90% and 10%) split scheme was used to train the classifier. Moreover, evaluation was performed with the gold standard expert labeled dataset to measure precision for each binary classified class.
Download
Paper Nr: 77
Title:

Design of a Personalized Affective Exergame to Increase Motivation in the Elderly
Authors:

Fenja T. Bruns and Frank Wallhoff
Abstract: Older people are among the most physically inactive. Game-based training programs (exergames) can motivate this group to exercise more. However, for long term benefit it is important to maintain engagement and motivation. Therefore, the emotions of the player should also be taken into account. This paper first gathers requirements to develop a motivating exergame for elderly people. This includes specifics regarding the target group as well as the inclusion of emotion theories. Based on these considerations, a concept for a framework is proposed how these requirements can be implemented with the help of personalization and sensor technology to create a successful and motivating exergame. A sample application demonstrates the flexibility of the presented framework.
Download
Paper Nr: 80
Title:

A Prognostic Machine Learning Framework and Algorithm for Predicting Long-term Behavioural Outcomes in Cancer Survivors
Authors:

Anneliese Markus, Amos Roche, Chun-Kit Ngan, Yin-Ting Cheung and Kristi Prifti
Abstract: We propose a prognostic machine learning (ML) framework to support the behavioural outcome prediction for cancer survivors. Specifically, our contributions are four-fold: (1) devise a data-driven, clinical domain-guided pipeline to select the best set of predictors among cancer treatments, chronic health conditions, and socio-environmental factors to perform behavioural outcome predictions; (2) use the state-of-the-art two-tier ensemble-based technique to select the best set of predictors for the downstream ML regressor constructions; (3) develop a StackNet Regressor Architecture (SRA) algorithm, i.e., an intelligent meta-modeling algorithm, to dynamically and automatically build an optimized multilayer ensemble-based RA from a given set of ML regressors to predict long-term behavioural outcomes; and (4) conduct a preliminarily experimental case study on our existing study data (i.e., 207 cancer survivors who suffered from either Osteogenic Sarcoma, Soft Tissue Sarcomas, or Acute Lymphoblastic Leukemia before the age of 18) collected by our investigators in a public hospital in Hong Kong. In this pilot study, we demonstrate that our approach outperforms the traditional statistical and computation methods, including Linear and non-Linear ML regressors.
Download
Paper Nr: 84
Title:

Multimodal Analysis of User-recipes Interactions
Authors:

Emilija Georgievska, Martina Stojanoska, Sanja Mishovska, Tome Eftimov and Dimitar Trajanov
Abstract: A good diet is essential for good health and nutrition, but also as a way of expressing and feeling good. Culinary and food recommender systems are becoming increasingly popular at a time when people are facing fast-paced lifestyles. In this paper, we are analysing interactions between users and recipes in order to make food recommendations based on their previous behaviour which would result in higher personalization for every single person. This also raises the question of whether people stick to what they know well or are open to new suggestions, or do personal recommendations lead to more homogeneity.
Download
Paper Nr: 86
Title:

Migration of Telemedicine Applications to National Telematics Infrastructure using Epilepsy Treatment as an Example
Authors:

Salima Houta, Tim Wilking, Marcel Klötgen and Falk Howar
Abstract: Digitization in epilepsy treatment, which usually is an intersectoral effort, offers great potential. Aggregated healthcare information from different actors involved in the treatment process provides an important basis for therapy decisions. More and more telemedicine solutions for the treatment of patients with epilepsy focus in particular on patient involvement via a digital seizure diary. This is intended to replace the currently mostly paper-based diaries. However, there is no widespread use in practice. The introduction of the national telematics infrastructure (TI) offers the opportunity to make telemedical applications accessible to a larger group of patients and medical institutions in Germany. The E-Health Act, which came into force on December 29, 2015, defines a roadmap for the gradual introduction of a telematics infrastructure in the German healthcare system. In addition to the specified TI components for secure and standardized data exchange, health IT service providers can migrate their existing digital solutions for healthcare in the TI. This article describes the migration of a developed telemedical infrastructure for epilepsy care into the national telematics infrastructure. First, an analysis of the telemedicine infrastructure is made with regard to supported integration options. Then, considering the chosen approach, an integration concept is designed using an example scenario.
Download
Paper Nr: 94
Title:

Digitization of Landmark Training for Topographical Disorientation: Opportunities of Smart Devices and Augmented Reality
Authors:

Tom Lorenz, Mirco Baseniak, Linda Münch, Ina Schiering and Sandra V. Müller
Abstract: Navigational abilities and wayfinding are important skills for participation in society. Landmark-based navigation is considered as an important basic wayfinding strategy. This strategy is used as the underlying concept for a rehabilitation training for people with topological disorientation. A digitization of this approach is proposed based on a smartphone application employing Augmented Realty concepts. This application allows to describe routes based on landmarks and a training of the defined routes. It is developed in an agile, interdisciplinary research process taking especially usability and privacy aspects into account.
Download
Paper Nr: 96
Title:

Local Explanations for Clinical Search Engine Results
Authors:

Edeline Contempré, Zoltán Szlávik, Majid Mohammadi, Erick Velazquez, Annette T. Teije and Ilaria Tiddi
Abstract: Health care professionals rely on treatment search engines to efficiently find adequate clinical trials and early access programs for their patients. However, doctors lose trust in the system if its underlying processes are unclear and unexplained. In this paper, a model-agnostic explainable method is developed to provide users with further information regarding the reasons why a clinical trial is retrieved in response to a query. To accomplish this, the engine generates features from clinical trials using by using a knowledge graph, clinical trial data and additional medical resources. Moreover, a crowd-sourcing methodology is used to determine features’ importance. Grounded on the proposed methodology, the rationale behind retrieving the clinical trials is explained in layman’s terms so that healthcare processionals can effortlessly perceive them. In addition, we compute an explainability score for each of the retrieved items, according to which the items can be ranked. The experiments validated by medical professionals suggest that the proposed methodology induces trust in targeted as well as in non-targeted users, and provide them with reliable explanations and ranking of retrieved items.
Download
Paper Nr: 5
Title:

Protecting Non-communicable Diseases Patients during Pandemics: Fundamental Rules for Engagement and the Case of Lebanon
Authors:

Luna El Bizri and Nabil G. Badr
Abstract: Non-communicable diseases (NCDs) are still the number one killer in the world. Their economic burden is heavy, notably in low-and-middle-income countries. Lebanon is a middle-income country in the Eastern Mediterranean region. The arising COVID-19 pandemic, political and economic instability, inadequate funding, and deteriorated infrastructure have rendered the country a fragile setting, significantly affecting persons with non-communicable diseases. Improving the patient journey during the COVID-19 pandemic and a comprehensive approach to NCD management is important during emergencies.This paper used a quantitate literature review to provide a theoretical framework touching NCDs patients in their journey during emergencies and crisis. It further adopted the Sendai Framework to draw the road for these patients in Lebanon. The ultimate goal is better preparedness and response in case of emergencies and disasters. It calls for a clear and coordinated action plan addressing the challenges posed by NCDs to a resilient country's response. This paper provides an overview of the situation of NCD patients in Lebanon during the COVID- 19 pandemic. It suggests strategies to address noncommunicable diseases guided by the Sendai Framework's four priorities, based on previous experiences.
Download
Paper Nr: 7
Title:

Disability Advocacy using a Smart Virtual Community
Authors:

Bushra Kundi, Dhayananth Dharmalingam, Rediet Tadesse, Alexandra Creighton, Rachel Gorman, Pierre Maret, Fabrice Muhlenbach, Alexis Buettgen, Enakshi Dua, Thumeka Mgwigwi, Serban Dinca-Panaitescu and Christo El Morr
Abstract: The lack of readily available disability data is a major barrier for disability advocacy globally. The collection and access to disability data is crucial to address social inequities, discrimination, and human rights violations within the disability community. The Disability Wiki project intends to use AI techniques such as Machine Learning and Semantic Web to extract and store existing disability-related data into one platform (Wikibase) and to provide a multilingual natural language enabled search engine and a screen-reader-accessible for its users.
Download
Paper Nr: 9
Title:

A Scoping Review of the Inquiry Instruments Being Used to Evaluate the Usability of Ambient Assisted Living Solutions
Authors:

Rute Bastardo, João Pavão and Nelson P. Rocha
Abstract: This paper reports a scoping review of the literature to identify the inquiry instruments being used to evaluate the usability of AAL solutions, which resulted in the inclusion of 35 studies. The results show that a significant number of the included studies reported the use of non-valid inquiry instruments, such as ad-hoc questionnaires. Among the studies using valid and reliable inquiry instruments, System Usability Scale (SUS) emerged as the most used one. In general, valid and reliable inquiry instruments are being used together with additional data gathering methods, to perform comprehensive usability evaluations. Moreover, in terms of the quality of the design of the included studies, it should be pointed the adequacy of the participants’ characteristics and the tasks they performed. In turn, these studies did not present evidence of the preparation and independence of the evaluators.
Download
Paper Nr: 17
Title:

Ensemble Feature Selection for Heart Disease Classification
Authors:

Houda Benhar, Ali Idri and Mohamed Hosni
Abstract: Feature selection is a fundamental data preparation task in any data mining objective. Deciding on the best feature selection technique to use for a specific context is difficult and time-consuming. Ensemble learning can alleviate this issue. Ensemble methods are based on the assumption that the aggregate results of a group of experts with average knowledge can often be superior to those of highly knowledgeable individual ones. The present study aims to propose a heterogeneous ensemble feature selection for heart disease classification. The proposed ensembles were constructed by combining the results of five univariate filter feature selection techniques using two aggregation methods. The performance of the proposed techniques was evaluated with four classifiers and six heart disease datasets. The empirical experiments showed that applying ensemble feature ranking produced very promising results compared to single ones and previous studies.
Download
Paper Nr: 21
Title:

Pima Indians Diabetes Database Processing through EBBM-Optimized UTM Model
Authors:

Luigi Lella, Ignazio Licata and Christian Pristipino
Abstract: A new predictive model tested on the Pima Indians Diabetes Database is presented. This model represents a particular subclass of A-Type Unorganized Turing machine (UTM), where the state is unique. It appears as a simple combinational network of NAND gates (it is not the more generic sequential type described by Turing, but it is enough to solve the examined predictive task). The optimal architecture of this network is identified by the use of evolutionary algorithms, which are therefore used as computational optimization algorithms. In particular, a classic genetic algorithm and an hybrid evolutionary-swarm algorithm that we have called Evolutionary Bait Balls Model (EBBM) were tested for this purpose. The predictive model thus defined, made it possible to achieve higher performances than those obtained with other classic predictive models. The final combinational network of NAND gates obtained through our model has allowed us to identify a simple Boolean rule to determine the existence of the risk of incurring diabetes mellitus.
Download
Paper Nr: 23
Title:

Benefits and Limitations of Age Group-Adjusted Average in the Profitability Audit for Pharmocotherapy
Authors:

Reinhard Schuster, Thomas Ostermann, Timo Emcke and Fabian Schuster
Abstract: Benchmarks for pharmaceuticals have been used for over 25 years to limit the cost increase in the second largest cost block in statutory health insurance in Germany with financial punishments for the physicians. The Regional Social Court of Dresden declares such a payback practice to be inadmissible if no age reference is used. In 2016, in most regions of the statutory health insurance associations, the division into status groups members, family members and pensioners has been changed into four age groups. The Supply Strengthening Act has opened up the possibility of drafting regional agreements. In Schleswig-Holstein, Morbidity Related Groups (MRG) were introduced for morbidity-related considerations. A number of other regions are currently using retrospective average cost limitations, which have the same problems as the benchmark restrictions. The aim of this paper is to investigate the influence of the type of health insurance (sickness) fund on the benchmark result with status and with age groups. Different morbidity structures between the health insurance funds are the subject of the risk structure compensation. For doctors, this aspect is not given sufficient consideration with respect to patient-specific morbidity characteristics till now.
Download
Paper Nr: 28
Title:

Enhancement of Physiological Stress Classification using Psychometric Features
Authors:

Karl Magtibay, Xavier Fernando and Karthikeyan Umapathy
Abstract: Psychological data features are underutilized in many acute stress studies since they are challenging to replicate and validate due to their inherent subjectivity. However, psychology and perception play essential roles in stress research according to the well-established allostatic load model. Therefore, we demonstrate the importance of accounting for psychological data in acute stress research in an ambulatory setting through a joint analysis. We enhanced stress classification by combining psychometric features with standard physiological signal features. We used the publicly available Wearable Stress and Affect Database (WESAD), from which we obtained physiological signals and psychological self-assessments from 15 participants. For each participant, a set of physiologically relevant features were extracted from each signal type. In parallel, we adapted psychometric features, positive emotion (PEscore) and negative emotion (NEscores) scores, by calculating the weighted average of self-evaluation scores. Using a stepwise feature selection and a linear- discriminant-analysis-based classifier, we found that PEscores, along with select physiological signal features, could enhance cross-validated stress classification accuracy by 8%, higher than a previous benchmark study using the same dataset. More importantly, we found that such a classification accuracy could be achieved with significantly fewer physiological signal features (by 20 times) with the aid of a psychometric feature. Finally, we found that psychometric features could indicate the type of perceived stress relating to an individual’s mood descriptor scores. Thus, a combination of psychometric and physiological data could be beneficial towards improving the detection and management of stress and support the development of holistic stress models.
Download
Paper Nr: 29
Title:

Towards an IoHT Platform to Monitor QoL Indicators
Authors:

Pedro M. Oliveira, Rossana C. Andrade, Pedro S. Neto and Breno S. Oliveira
Abstract: The Quality of Life has been studied for a long time, and the World Health Organization defines it as the individual perception about life regarding four major domains: physical, psychological, social, and environmental. The relevance to study QoL lies in the search for strategies able to measure a patient’s well-being. Without these strategies, treatments, and technological solutions that aim to improve people’s QoL would be restricted to physicians’ implicit and subjective perceptions. Thus, there are many instruments for formal QoL assessment (usually questionnaires). However, the use of these instruments is time-consuming, non-transparent, and error-prone. Considering this problem, in this work, we discuss the proposal to use the Internet of Health Things (IoHT) to collect data from smart environments and apply machine learning techniques to infer QoL measures. To achieve this goal, we designed an IoHT platform inspired by the MAPE-K loop. Our literature review has shown that this idea is promising and that there are many open challenges to be addressed.
Download
Paper Nr: 43
Title:

Study of the User Behaviour Caused by Automatic Recommendation Systems Call to Action
Authors:

Georgy Kopanitsa and Sergey Kovalchuk
Abstract: Diagnostics accuracy and usability of symptom checkers have been researched in several studies. Their ability to set a correct diagnosis especially in the urgent cases is questionable. There is one aspect of symptom checkers that has not been deeply studied yet. It is their ability to motivate patients to follow up after receiving a direct recommendation and to decrease a load on the health care professionals. The goal of this research is to study how patients behave after receiving a recommendation from a symptom checker and motivation of this behaviour. We studied how patients react on the symptom checker recommendations and the motivation behind this behaviour. In total we invited 3615 patients to have a symptom checker screening; 2374 of them agreed to run a symptom checker screening; 867 of them agreed to participate in the study. The proportion of the patients who agreed to have a symptom checker screening. So, we can clearly see that symptom checker screening doesn’t result in a significant decrease of the load on healthcare professionals. This is supported by the quantitative study results. The patients emphasized the ease of use of the tool and clearness of the recommendations it gives. However, they perceived it as rather a second opinion tool or a tool that helps to prepare to the doctor’s visit.
Download
Paper Nr: 53
Title:

Mobile Applications for Self-management of Chronic Diseases: A Systematic Review
Authors:

Benjamin Stahr, Sebastian Fudickar and Christian Lins
Abstract: Objectives: Since apps have been gaining popularity, they are also used to support the treatment of chronic diseases. However, the effectiveness of these measures has not been fully confirmed. This review deals with features that make these apps effective. Methods: In this structured literature survey, relevant studies from the year 2014 to 2019 were identified. Inclusion criteria were that the study included an app that was used to alleviate symptoms of chronic diseases or was intended to support the preventive treatment of patients. Results: Ten studies were examined in detail, of which seven found significant effects. Factors, which increase the effectiveness of mHealth apps include easy integration into everyday life, appropriate training of users, tailoring the app to the target group, focusing on improving the relationship between user and disease, and user-specific treatment of symptoms. Tracking of symptoms, education, and a chat can also increase effectiveness. Conclusions: Most of the papers reviewed showed a positive impact of mobile apps on chronic disease progression. However, a negative factor was also identified, in which patients became more involved with their illnesses as a result of the intervention, which increased the perceived severity of the illness and thus reduced the quality of life.
Download
Paper Nr: 54
Title:

Assessing Occupational Health with a Cross-platform Application based on Self-reports and Biosignals
Authors:

Sara Silva, Catia Cepeda, João Rodrigues, Phillip Probst and Hugo Gamboa
Abstract: Occupational disorders have a significant impact on the health of office workers. This has even more relevance considering the increased population in this work modality and the recent shift to remote work. Efforts are needed to create worker awareness and reduce occupational hazards. Based on this motivation, an intuitive and easy to use application for the assessment of occupational risks was developed and it is presented in this paper. This application records risk factors in the biomechanical, psycho-social, and environmental domains through data collected with self-assessment tools and wearable sensors, contributing to a more complete, robust and personal assessment of risk exposure. This article presents the system architecture and its application interface. Examples of interaction with each module of the app are also provided.
Download
Paper Nr: 57
Title:

Classifying Breast Cytological Images using Deep Learning Architectures
Authors:

Hasnae Zerouaoui and Ali Idri
Abstract: Breast cancer (BC) is a leading cause of death among women worldwide. It remains a critical challenge, causing over 10 million deaths globally in 2020. Medical images analysis is the most promising research area since it provides facilities for diagnosing several diseases such as breast cancer. The present paper carries out an empirical evaluation of recent deep Convolutional Neural Network (CNN) architectures for a binary classification of breast cytological images based fined tuned versions of seven deep learning techniques: VGG16, VGG19, DenseNet201, InceptionResNetV2, InceptionV3, ResNet50 and MobileNetV2. The empirical evaluations used: (1) four classification performance criteria (accuracy, recall, precision and F1-score), (2) Scott Knott (SK) statistical test to select the best cluster of the outperforming architectures, and (3) borda count voting system to rank the best performing architectures. All the evaluations were over the FNAC dataset which contain 212 images. Results showed the potential of deep learning techniques to classify breast cancer in malignant and benign, therefor the findings of this study recommend the use of MobileNetV2 for the classification of the breast cancer cytological images since it gave the best results with an accuracy of 98.54%.
Download
Paper Nr: 62
Title:

The Mindfulness Meditation Effect on States of Anxiety, Depression, Stress and Quality of Life
Authors:

Pedro Morais, Ana P. Pinheiro, Miguel S. Fonseca and Carla Quintão
Abstract: Purpose: This paper aims to study how mindfulness meditation can be used to prevent, or improve, states of anxiety, depression, stress and loss of quality of life. Although this model of meditation has been associated with a healthier life, there is a need for scientific evidence-based on longitudinal results. Methods: Twenty-five volunteers, asymptomatic of psychological distress, participated in this research project attending a Mindfulness-Based Stress Reduction (MBSR) course. The status of each individual was assessed for 18 weeks, with three scales: World Health Organization Quality of Life (WHOQOL), Profile of Mood States (POMS) and Depression, Anxiety and Stress Scale (DASS). There were four evaluation periods: Pre/Peri/Post-MBSR course and a fourth follow-up, after two months. Results: Comparing the beginning to the end of the MBSR course, a significant reduction was observed in mean results of self-reported anxiety: -66.0% (p<0.001), stress: -52.0% (p<0.001), depression: -51.0% (p<0.001) and Total Mood Disturbance (TMD): -19.0% (p<0.001), as well as an increase in quality of life: 11.2% (p<0.001). Conclusion: The current values suggest that the practice of mindfulness meditation, characterized by self-regulation of attention, can be used as a proactive way to prevent and respond to psychopathological disorders.
Download
Paper Nr: 68
Title:

Emergency Health Protocols Supporting Health Data Exchange, Cloud Storage, and Indexing
Authors:

Konstantinos Koutsoukos, Chrysostomos Symvoulidis, Athanasios Kiourtis, Argyro Mavrogiorgou, Stella Dimopoulou and Dimosthenis Kyriazis
Abstract: The health industry has evolved significantly through the last years by adapting to the new technologies and exploiting them in order to upgrade the services that provides to the people. In this context, a lot of effort has been focused on converting medical documents to electronic health records and storing them online. However, taking into consideration the current innovations, it is doubtless that there are many limitations when these proposals are applied in a real-life scenario. For this reason, this paper proposes a system that combines electronic data storage and health record exchange between individuals and authenticated medical staff in a secure way. The specific recommendation is being evaluated through the corresponding applications and protocols that are developed and finally, the results exhibit the solutions over existing gaps.
Download
Paper Nr: 79
Title:

STREAM: Prototype Development of a Digital Language Identifier
Authors:

Rebecca Meehan, Manisha Kumari, Qiang Guan, Sanda Katila, Joel Davidson and Nichole Egbert
Abstract: This paper describes the development of a prototype digital language identifier (STREAMTM), designed to help improve health by shortening the time it takes for healthcare professionals and first responders working with non-native speakers to identify a patient’s language so that the correct interpreter can be called, enabling the patient to more quickly get the care they need. The innovation was originally designed to address the needs for health care professionals caring for the emergent Bhutanese and Nepali community of newcomers in Akron, Ohio, USA. Language access support through in person and electronic interpreters continues to improve, however, there remains a need to quickly identify spoken language at points of entry or in emergent situations. We developed and tested a digital prototype solution (Smart Translation Enabling and Aiding Multicultural populations, aka “STREAMTM” tool, patent pending) based on the Nepali language that can be later extended to identify multiple languages. Prototype testing of STREAMTM showed that although the model predicted the correct language better than chance, accuracy needs to be improved. Next steps include refining the model to identify spoken language in a shorter amount of time, adding other languages to the model, and user testing among medical and emergency services professionals.
Download
Paper Nr: 83
Title:

How Long Are Various Types of Daily Activities? Statistical Analysis of a Multimodal Wearable Sensor-based Human Activity Dataset
Authors:

Hui Liu and Tanja Schultz
Abstract: Human activity research in the field of informatics, such as activity segmentation, modeling, and recognition, is moving in an increasingly interpretable direction with the introduction of sports and kinematics knowledge. Many related research topics face a question: How long is the typical duration of the activities needed to be modeled? Several public human activity datasets do not strictly limit single motions’ repetition times, such as gait cycle numbers, in recording sessions, so they are not statistically significant concerning activity duration. Standing on the rigorous acquisition protocol design and well-segmented data corpus of the recently released multimodal wearable sensor-based human activity dataset CSL-SHARE, this paper analyzes the duration statistics and distribution of 22 basic single motions of daily activities and sports, hoping to provide research references for human activity studies. We discovered that (1) the duration of each studied human daily activity or simple sports activity reflects interpersonal similarities and naturally obeys a normal distribution; (2) one single motion (such as jumping and sitting down) or one cycle in the activities of cyclical motions (such as one gait cycle in walking) has an average duration in the interval from about 1 second to 2 seconds.
Download
Paper Nr: 89
Title:

Privacy Notifications for Transparency in Fitness Apps
Authors:

Mirco Baseniak, Tom Lorenz and Ina Schiering
Abstract: mHealth applications including fitness apps are an important trend. To monitor fitness activities a broad range of personal data is processed typically including location data and vital signs. For some of these applications it is not transparent which data is processed. To foster transparency and intervenability in mobile applications the concept of privacy notifications is an opportunity to provide users with information about processed data during the use of the application. In the context of a fitness app a concept for privacy notifications is proposed and evaluated in a user study.
Download
Paper Nr: 91
Title:

Discussion on Comparing Machine Learning Models for Health Outcome Prediction
Authors:

Janusz Wojtusiak and Negin Asadzadehzanjani
Abstract: This position paper argues the need for more details than simple statistical accuracy measures when comparing machine learning models constructed for patient outcome prediction. First, statistical accuracy measures are briefly discussed, including AROC, APRC, predictive accuracy, precision, recall, and their variants. Then, model correlation plots are introduced that compare outputs from two models. Finally, a more detailed analysis of inputs to the models is presented. The discussions are illustrated with two classification problems in predicting patient mortality and high utilization of medical services.
Download
Paper Nr: 92
Title:

A Systematic Map of Interpretability in Medicine
Authors:

Hajar Hakkoum, Ibtissam Abnane and Ali Idri
Abstract: Machine learning (ML) has been rapidly growing, mainly owing to the availability of historical datasets and advanced computational power. This growth is still facing a set of challenges, such as the interpretability of ML models. In particular, in the medical field, interpretability is a real bottleneck to the use of ML by physicians. This review was carried out according to the well-known systematic map process to analyse the literature on interpretability techniques when applied in the medical field with regard to different aspects. A total of 179 articles (1994-2020) were selected from six digital libraries. The results showed that the number of studies dealing with interpretability increased over the years with a dominance of solution proposals and experiment-based empirical type. Additionally, artificial neural networks were the most widely used ML black-box techniques investigated for interpretability.
Download
Paper Nr: 97
Title:

Opportunities for System Dynamics towards the Support of Technological Developments in Stroke Treatment Domain
Authors:

Julia Kantorovich and Jukka Ranta
Abstract: Data driven solutions can facilitate and enhance stroke diagnostics and at the same time management of stroke prevention and treatment in a cost-effective way. However, the potential and the utilization of data and AI analytics in stroke solutions are largely neglected. At the same time, the process to enter to medical domain for technology developer is not straightforward. There is a need for common vocabularies and design tools to engage medical professionals in interaction with technologists during the research and development phase to let them know what is needed. This paper valorises the opportunities for System Dynamics to support technology developers in the developing of innovative solutions and applications for stroke diagnosis and treatment. In addition, the value of System Dynamics to support the impact analysis (health outcome, decision quality, care costs, etc.) and hereby to facilitate the business and market uptake of new innovative solutions in this domain is demonstrated.
Download