HEALTHINF 2021 Abstracts


Full Papers
Paper Nr: 2
Title:

Assessing Postures and Mechanical Loads during Patient Transfers

Authors:

Sandra Hellmers, Anna Brinkmann, Conrad F. Böhlen, Sandra Lau, Rebecca Diekmann and Andreas Hein

Abstract: Socio-Demographic developments in industrialized countries cause a discrepancy between potential recipients and providers of care. Caregivers experience high musculoskeletal loads during their daily work, which leads to back complaints and a high rate of absenteeism at work. Ergonomically correct working can significantly reduce musculoskeletal load. In a study with 13 caregiver students, we analyzed body postures, muscle activities, and loads during the transfer of a patient from bed to wheelchair. Our measurement system consists of a full-body motion capture system and a Multi Kinect System. Additionally, muscle activities were measured via surface electromyography. According to recommendations for ergonomic working in the care sector, a system was developed that recognizes potentially harmful postures based on the motion capture data. A result report visualizes the skeleton model together with color-coded information about inclination and torsion angles. The motion capture data was also related to EMG data and analyzed according to biomechanical assumptions.

Paper Nr: 3
Title:

A Study about Discovery of Critical Food Consumption Patterns Linked with Lifestyle Diseases for Swiss Population using Data Mining Methods

Authors:

Ilona R. Mewes, Helena Jenzer and Farshideh Einsele

Abstract: Background: This article demonstrates that using data mining methods such as association analysis on an integrated Swiss database derived from a Swiss national dietary survey (menuCH) and Swiss demographical and health data is a powerful way to determine whether a specific population subgroup is at particular risk for developing a lifestyle disease based on its food consumption patterns. Objective: The objective of the study was to use an integrated database of dietary and health data from a large group of Swiss population to discover critical food consumption patterns linked with lifestyle diseases known to be strongly tied with food consumption. Design: Food consumption databases from a Swiss national survey menuCH were gathered along with corresponding large survey of demographics and health data from Swiss population conducted by Swiss Federal Office of Public Health (FOPH). These databases were integrated and reported in a previous study as a single integrated database. A data mining method such as A-priori association analysis was applied to this integrated database. Results: Association mining analysis was used to incorporate rules about food consumption and lifestyle diseases. A set of promising preliminary rules and their corresponding interpretation was generated, which is reported in this paper. As an example, the found rules of the sample show that smoking is relatively irrelevant to the high blood pressure and Diabetes, whereas consuming vegetables at regular basis reduces the risk of high Cholesterol. Conclusions: Association rule mining was successfully used to describe and predict rules linking food consumption patterns with lifestyle diseases. The gained association rules reveal that the appearance of the mutually independent nutritional characteristics in the rules are equally distributed. Furthermore, most of the sample show no chronical diseases as they smoke little and exercise regularly, which can be interpreted that sport is a strong preventive factor for chronical/lifestyle diseases. Nevertheless, a small percentage of the sample shows chronic illnesses due to unhealthy eating. Further research should consider the weighting of chronic diseases’ characteristics for them not to be pruned out early by data mining computation.

Paper Nr: 9
Title:

Evaluating a Multi Depth Camera System to Consolidate Ergonomic Work in the Education of Caregivers

Authors:

Conrad F. Böhlen, Anna Brinkmann, Sebastian Fudickar, Sandra Hellmers and Andreas Hein

Abstract: Through the demographic change in Western European countries the demand for nurses in elderly care rises. Additionally, constant high physical stresses causing nurses to leave the profession before the retirement age and having a high number of sick leave days. To reduce the physical strain, focus must be set to ergonomically correct work in nursing schools. Currently given technical infrastructure in the schools lacks the capability to provide nursing instructors with analyzable data from simulated care acts. In this work, we present and evaluate the Multi-Kinect-System, our custom developed depth sensor system for recording and analyzing care acts. In a study, 13 students of a nursing school performed a simulated transfer tasks under observation of a nursing instructor and our system. The instructor gives a more in-depth evaluation of the transfer when using the intuitively analyzable data of our system, regarding the feedback length and information content.

Paper Nr: 11
Title:

Terminology Expansion with Prototype Embeddings: Extracting Symptoms of Urinary Tract Infection from Clinical Text

Authors:

Mahbub U. Alam, Aron Henriksson, Hideyuki Tanushi, Emil Thiman, Pontus Naucler and Hercules Dalianis

Abstract: Many natural language processing applications rely on the availability of domain-specific terminologies containing synonyms. To that end, semi-automatic methods for extracting additional synonyms of a given concept from corpora are useful, especially in low-resource domains and noisy genres such as clinical text, where nonstandard language use and misspellings are prevalent. In this study, prototype embeddings based on seed words were used to create representations for (i) specific urinary tract infection (UTI) symptoms and (ii) UTI symptoms in general. Four word embedding methods and two phrase detection methods were evaluated using clinical data from Karolinska University Hospital. It is shown that prototype embeddings can effectively capture semantic information related to UTI symptoms. Using prototype embeddings for specific UTI symptoms led to the extraction of more symptom terms compared to using prototype embeddings for UTI symptoms in general. Overall, 142 additional UTI symptom terms were identified, yielding a more than 100% increment compared to the initial seed set. The mean average precision across all UTI symptoms was 0.51, and as high as 0.86 for one specific UTI symptom. This study provides an effective and cost-effective solution to terminology expansion with small amounts of labeled data.

Paper Nr: 15
Title:

Detecting Dyslexia from Audio Records: An AI Approach

Authors:

Jim Radford, Gilles Richard, Hugo Richard and Mathieu Serrurier

Abstract: Dyslexia impacts the individual’s ability to read, interferes with academic achievements and may also have long term consequences beyond the learning years. Early detection is critical. It is usually done via a lengthy battery of tests: human experts score these tests to decide whether the child requires specific education strategies. This human assessment can also lead to inconsistencies. That is why there is a strong need for earlier, simpler (and cheaper) screening of dyslexia. In this paper, we investigate the potential of modern Artificial Intelligence in automating this screening. With this aim in mind and building upon previous works, we have gathered a dataset of audio recordings, from both non-dyslexic and dyslexic children. After proper preprocessing, we have applied diverse machine learning algorithms in order to check if some hidden patterns are discoverable, making a difference between dyslexic and non-dyslexic readers. Then, we built up our own neural network which outperforms the other tested approaches. Our results suggests the possibility to classify audio records as characteristic of dyslexia, leading to an accurate and inexpensive dyslexia screening via non-invasive methods, potentially reaching a large population for early intervention.

Paper Nr: 18
Title:

Exploiting Food Embeddings for Ingredient Substitution

Authors:

Chantal Pellegrini, Ege Özsoy, Monika Wintergerst and Georg Groh

Abstract: Identifying ingredient substitutes for cooking recipes can be beneficial for various goals, such as nutrient optimization or avoiding allergens. Natural language processing (NLP) techniques can be valuable tools to make use of the vast cooking-related knowledge available online, and aid in finding ingredient alternatives. Despite previous approaches to identify ingredient substitutes, there is still a lack of research in this area regarding the most recent developments in the field of NLP. On top of that, a lack of standardized evaluation metrics makes comparing approaches difficult. In this paper, we present two models for ingredient embeddings, Food2Vec and FoodBERT. In addition, we combine both approaches with images, resulting in two multimodal representation models. FoodBERT is furthermore used for relation extraction. We conduct a ground truth based evaluation for all approaches, as well as a human evaluation. The comparison shows that FoodBERT, and especially the multimodal version, is best suited for substitute recommendations in dietary use cases.

Paper Nr: 20
Title:

Comparison of Algorithms to Measure a Psychophysical Threshold using Digital Applications: The Stereoacuity Case Study

Authors:

Silvia Bonfanti and Angelo Gargantini

Abstract: The use of digital applications to perform psychophysical measurements led to the introduction of algorithms to guide the users in test execution. In this paper we show three algorithms, two already well known: StrictStaircase and PEST, and a new one that we propose: PEST3. All the algorithms aim at estimating the level of a psychophysical capability by performing a sequence of simple tests; starting from initial level N, the test is executed until the target level is reached. They differ in the choice of the next steps in the sequences and the stopping condition. We have applied the algorithms to the stereoacuity case study and we have compared them by answering a set of research questions. Finally, we provide guidelines to choose the best algorithm based on the test goal. We found that while StrictStaircase provides optimal results, it requires the largest number of steps and this may hinder its use; PEST3 can overcome these limits without compromising the final results.

Paper Nr: 22
Title:

Developing a Machine Learning Workflow to Explain Black-box Models for Alzheimer’s Disease Classification

Authors:

Louise Bloch and Christoph M. Friedrich

Abstract: Many research articles used difficult-to-interpret black-box Machine Learning (ML) models to classify Alzheimer’s disease (AD) without examining their biological relevance. In this article, an ML workflow was developed to interpret black-box models based on Shapley values. This workflow enabled the model-agnostic visualization of complex relationships between model features and predictions and also the explanation of individual predictions, which is important in clinical practice. To demonstrate this workflow, eXtreme Gradient Boosting (XGBoost) and Random Forest (RF) classifiers were trained for AD classification. All models were trained on the Alzheimer’s Disease Neuroimaging Initiative (ADNI) or Australian Imaging and Lifestyle flagship study of Ageing (AIBL) dataset and were validated for independent test datasets of both cohorts. The results showed improved performances for black-box models in comparison to simple Classification and Regression Trees (CARTs). For the classification of Mild Cognitive Impairment (MCI) conversion and the ADNI training dataset, the best model achieved a classification accuracy of 71.03 % for the ADNI test dataset and 67.65 % for the entire AIBL dataset. This RF used a logical long-term memory test, the count of Apolipoprotein E ε4 (ApoEε4) alleles and the volume of the left hippocampus as the most important features.

Paper Nr: 27
Title:

Practical Application of Maturity Models in Healthcare: Findings from Multiple Digitalization Case Studies

Authors:

Anja Burmann and Sven Meister

Abstract: Maturity models (MMs) are widely applied means for describing a current status of development across numerous domains, but also within the healthcare sector. They offer orientation for systematic development and improvement regarding the aspect examined. However, experience from the practical application of MMs is scarcely described. Within this article two projects in which MMs were used in collaboration with practitioner groups from hospital environments are presented. The general project intentions, motivation for incorporating MMs and generated results are described. By deriving observations across the two cases, general tensions between healthcare practice and research concerning maturity modelling are identified. Additionally, the suitability of existing MMs to support especially hospitals in coping with the challenges of the digital transformation is discussed. This study’s findings may be incorporated into development and refinement of MMs and may thus contribute to increasing practical value created by such means.

Paper Nr: 30
Title:

Behavioral Predictive Analytics towards Personalization for Self-management

Authors:

Bon Sy, Jin Chen, Magdalen Beiting-Parrish and Connor Brown

Abstract: The objective of this research is to investigate the feasibility of applying behavioral predictive analytics to optimize diabetes self-management. In the U.S., less than 25% of patients actively engage in self-management even though self-management has been reported to associate with improved health outcomes and reduced healthcare costs. The proposed behavioral predictive analytics relies on manifold clustering to derive nonlinear clusters. These clusters are characterized by behavior readiness patterns for subpopulation segmentation. For each subpopulation, an individualized auto-regression model and a population-based model are developed to support self-management personalization in three areas: glucose self-monitoring, diet management, and exercise. The goal is to predict personalized activities that are most likely to achieve optimal engagement. This paper reports the result of manifold clusters based on 148 subjects with type 2 diabetes, and shows the preliminary result of personalization for 22 subjects under different scenarios.

Paper Nr: 32
Title:

Automated Mobile Image Acquisition of Macroscopic Dermatological Lesions

Authors:

Dinis Moreira, Pedro Alves, Francisco Veiga, Luís Rosado and Maria M. Vasconcelos

Abstract: The incidence of skin cancer has been rising every year translating in high economic costs. The development of mobile teledermatology applications that can contribute for the standardization of image acquisition can facilitate early diagnosis. This paper presents a new methodology for real-time automated image acquisition of macroscopic skin images via mobile devices. It merges an automated image focus assessment that uses a feature-based machine learning approach with segmentation of dermatological lesions using computer vision techniques. It also describes the datasets used to develop and evaluate the proposed approach: 3428 images from one dataset purposely collected using different mobile devices for the focus assessment component, and a total of 1380 images from two other datasets available on the literature to develop the segmentation approach. The best model for automatic focus assessment of preview images and acquired picture achieved an overall accuracy of 88.3% and 86.8%, respectively. The segmentation approach attained a Jaccard index of 85.81% and 68.59% for SMARTSKINS and Dermofit datasets, respectively. The developed algorithms present a fast processing time that is suitable for real-time usage in medium and high performance smartphones. These findings were also validated by implementing the proposed methodology within an android application demonstrating promising results.

Paper Nr: 35
Title:

An Extension of Chronicles Temporal Model with Taxonomies: Application to Epidemiological Studies

Authors:

Johanne Bakalara, Thomas Guyet, Olivier Dameron, André Happe and Emmanuel Oger

Abstract: Medico-administrative databases contain information about patients’ medical events, i.e. their care trajectories. Semantic Web technologies are used by epidemiologists to query these databases in order to identify patients whose care trajectories conform to some criteria. In this article we are interested in care trajectories involving temporal constraints. In such cases, Semantic Web tools lack computational efficiency while temporal pattern matching algorithms are efficient but lack of expressiveness. We propose to use a temporal pattern called chronicles to represent temporal constraints on care trajectories. We also propose an hybrid approach, combining the expressiveness of SPARQL and the efficiency of chronicle recognition to query care trajectories. We evaluate our approach on synthetic data and real large data. The results show that the hybrid approach is more efficient than pure SPARQL, and validate the interest of our tool to detect patients having venous thromboembolism disease in the French medico-administrative database.

Paper Nr: 36
Title:

Automated Model for Tracking COVID-19 Infected Cases till Final Diagnosis

Authors:

Mohamed A. Gomaa, Mustafa Wassel, Rouzan M. Abdelmawla, Nihal Ibrahim, Khaled Nasser, Nermin A. Osman and Walid Gomaa

Abstract: The COVID-19 pandemic is now devastating. It affects public safety and well-being. A crucial step in the COVID-19 battle will be tracking the positive cases with convenient accuracy of diagnosis. However, the time of pandemics shows the emergent need for automated diagnosis to support medical staff decisions in different steps of diagnosis and prognosis of target disease like medical imaging through X-rays, CT-Scans, etc. Besides laboratory investigation steps, we propose a system that provides an automated multi-stage decision system supported with decision causes using deep learning techniques for tracking cases of a target disease (COVID-19 in our paper). Encouraged by the open-source Data sets for COVID-19 infected patients’ chest radiology, we proposed a system of three Consecutive stages. Each stage consists of a deep learning binary classifier tailored for the detection of a specific COVID-19 infection feature from chest radiology, either X-ray or CT-scan. By integrating the three classifiers, a multi-stage diagnostic system was attained that achieves an accuracy of (87.980 %), (78.717%), and (84%) for the three stages, respectively. By no means a production- ready solution, our system will help in reducing errors caused by human decisions, taken under pressure, and exhausting routines, and it will be reliable to take urgent decisions once the model performance achieves the needed accuracy.

Paper Nr: 40
Title:

Analysis of School Performance of Children and Adolescents with Attention-Deficit/Hyperactivity Disorder: A Dimensionality Reduction Approach

Authors:

Caroline Jandre, Bruno Santos, Marcelo Balbino, Débora de Miranda, Luis Zárate and Cristiane Nobre

Abstract: Attention-Deficit/Hyperactivity Disorder (ADHD) is defined by harmful inattention, disorganization, and/or hyperactivity and impulsivity. ADHD can negatively affect an individual’s life, but it is not a definitive factor for poor school performance. This work aims to identify classification rules that best describe the school performance in arithmetic, writing, and reading of students with ADHD. For this, information obtained from the Genetic Algorithm, Random Forest and specialists in ADHD were used so that later the VTJ48 and JRip algorithms could be applied. It is usual in the health area to collect various information about the individual, resulting in the frequent need to reduce the base’s dimensionality. The results found were promising, reaching up to 92% of F-Measure. The discovered rules point to environmental and emotional factors as drivers of school performance prognosis and reinforce that ADHD is not synonymous with academic failure.

Paper Nr: 44
Title:

Historical Report Assist Medical Report Generation

Authors:

Shan Ye, Mei Wang and Yijie Dong

Abstract: How to automatically generate diagnostic reports with accurate content, standardized structure and clear semantics, brings great challenges due to the complexity of medical images and the detailed paragraph descriptions for medical images. The structure and the semantic contents of the historical report are very helpful for the current report generation. This paper proposes a text report generation method assisted by historical reports. In the proposed method, both the previous report and the keywords generated from the current images are modeled by using two encoders respectively. The co-attention mechanism is introduced to jointly learn the historical reports and the keywords. The decoder based on the co-attention is used to generate a long description of the image. The progress that learns from the historical report and the current report in the training set helps to generate an accurate report for the new image. Furthermore, the structure in the historical report helps to generate a more natural text report. We conducted experiments on the practical ultrasound data, which is provided by a prestigious hospital in China. The experimental results show that the reports generated by the proposed method are closer to the reports generated by radiologists.

Paper Nr: 49
Title:

Challenges in Developing Software for the Swedish Healthcare Sector

Authors:

Bilal Maqbool and Sebastian Herold

Abstract: Context: High-quality software is essential to the progressing digitalisation of the Swedish healthcare sector. Developing software with the desired high quality is far from trivial due to the sophisticated requirements of the domain. Problem: Studies on healthcare digitalisation challenges in Sweden and other countries, however, largely focus on the perceptions of healthcare professionals and patients and less on opinions of IT professionals. Method: In this exploratory study, we conducted semi-structured interviews with nine IT professionals about observed challenges in developing software for the Swedish healthcare sector. A qualitative analysis was performed to identify common themes. Results: We identified the prevalent challenges to be related to data integrity, privacy and security, rules and regulations, engineering usability, and software testing. Conclusion: The results suggest that further research is required regarding agile methods, efficient requirement engineering, and testing in eHealth as well as in privacy and usability engineering.

Paper Nr: 56
Title:

Enhancing Cross-lingual Semantic Annotations using Deep Network Sentence Embeddings

Authors:

Ying-Chi Lin, Phillip Hoffmann and Erhard Rahm

Abstract: Annotating documents using concepts of ontologies enhances data quality and interoperability. Such semantic annotations also facilitate the comparison of multiple studies and even cross-lingual results. The FDA therefore requires that all submitted medical forms have to be annotated. In this work we aim at annotating medical forms in German. These standardized forms are used in health care practice and biomedical research and are translated/adapted to various languages. We focus on annotations that cover the whole question in the form as required by the FDA. We need to map these non-English questions to English concepts as many of these concepts do not exist in other languages. Due to the process of translation and adaptation, the corresponding non-English forms deviate from the original forms syntactically. This causes the conventional string matching methods to produce low annotation quality results. Consequently, we propose a new approach that incorporates semantics into the mapping procedure. By utilizing sentence embeddings generated by deep networks in the cross-lingual annotation process, we achieve a recall of 84.62%. This is an improvement of 134% compared to conventional string matching. Likewise, we also achieve an improvement of 51% in precision and 65% in F-measure.

Paper Nr: 58
Title:

Establishing the State of Practice about Data Standards in Monitoring Healthcare Interventions for HIV in Uganda’s EMR-based Health Information Systems

Authors:

Achilles Kiwanuka, Moses Bagyendera, Joseph Wamema, Andrew Alunyu, Mercy Amiyo, Andrew Kambugu and Josephine Nabukenya

Abstract: Electronic Health Information Systems (EHIS) in Uganda are characterised by inaccessibility to reliable, timely and integrated data for effectively monitoring and tracking continuity of care for people living with HIV, exacerbated by disparate, fragmented EHIS in varying health system levels that are not interoperable and lack common data standards. In order for data to be comparable, there has to be uniformity in terms of standards that are employed in a uniform manner in all data management processes. In this study, we established the state of current practice regarding data and interoperability standards in monitoring and evaluating healthcare interventions for HIV in Uganda’s EMR-based health information systems. The study findings indicate that there are scanty practices and/or implementation of the eHealth standards (data and interoperability), and limited to noncompliance of monitoring these standards in the implementation of the HIV healthcare interventions. Accordingly, our study recommendations point to the need of designing data and interoperability frameworks to provide for the specific set of standards, protocols, procedures, best practices and policies for eHealth standardisation in Uganda’s health system.

Paper Nr: 63
Title:

Automatic Real-time Beat-to-beat Detection of Arrhythmia Conditions

Authors:

Giovanni Rosa, Gennaro Laudato, Angela R. Colavita, Simone Scalabrino and Rocco Oliveto

Abstract: With the spread of Internet of Medical Things (IoMT) systems, the scientific community has dedicated a lot of effort in the definition of approaches for supporting specialized staff in the early diagnosis of pathological conditions and diseases. Several approaches have been defined for the identification of arrhythmia, a pathological condition that can be detected from an electrocardiogram (ECG) trace. There exist many types of arrhythmia and some of them present a great impact on the patients in terms of worsening of physical conditions or even mortality. In this work we present NEAPOLIS, a novel approach for the accurate detection of arrhythmia conditions. NEAPOLIS takes as input a heartbeat signal, extracted from an ECG trace, and provides as output a 5-class classification of the beat, namely normal sinus rhythm and four main types of arrhythmia conditions. NEAPOLIS is based on ECG characteristics that do not need a long-term observation of an ECG for the classification of the beat. This choice makes NEAPOLIS a (near) real-time detector of arrhythmia because it allows the detection within few seconds of ECG observation. The accuracy of NEAPOLIS has been compared to one of the best and most recent work from the literature. The achieved results show that NEAPOLIS provides a more accurate detection of arrhythmia conditions.

Paper Nr: 67
Title:

A Meta-ontology Framework for Parameter Concepts of Disease Spread Simulation Models

Authors:

Le Nguyen and Deborah Stacey

Abstract: This work reports on an ontological organization (framework) that separates domain knowledge from knowledge of specific views and formalizes conceptual relationships by linking to the meta-ontology structure. We use parameters of animal disease spread simulation models as an example, although all concepts presented could apply to human disease spread simulation as well. A meta-ontology is created to document parameter concepts in different comparable simulation models. It formalizes relationships between parameter concepts. This offers several advantages such as allowing explicit domain knowledge representation and provenance, allowing for the assessment of parameters with respect to domain knowledge, and assisting in usage and evaluation of the models. The meta-ontology allows views about parameter concepts to be captured. This is important because it establishes a neutral view point which allows the assessment of parameter semantics in respect to documented domain knowledge. While this work uses the domain of animal disease spread, the principles of ontological representation of model parameters is applicable to a wide range of domains.

Paper Nr: 77
Title:

Deep Learning Assisted Retinopathy of Prematurity Screening Technique

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 preterm babies worldwide. By using proper scanning and treatment, the effect of the blindness of ROP can be reduced. However, due to lack of medical facilities, a large proportion of these preterm infants remain undiagnosed after birth. As a result, these babies are more likely to have ROP induced blindness. In this paper, we propose a robust and intelligent system based on deep learning and computer vision to automatically detect the optical disk (OD) and retinal blood vessels and also classify the high severity (Zone-1) case of ROP. To test and validate the proposed system, we present empirical results using the preterm infant fundus images from a local hospital. Our results showed that the YOLO-V5 model accurately detects the OD from preterm babies fundus images. Further, the computer vision-based system accurately segmented the retinal vessels from the preterm babies fundus images. Specifically for the Zone-1 case of ROP, our system is able to achieve an accuracy of 83.3%.

Paper Nr: 94
Title:

Automatic Brain White Matter Hypertinsities Segmentation using Deep Learning Techniques

Authors:

José A. Viteri, Francis R. Loayza, Enrique Pelaez and Fabricio Layedra

Abstract: White Matter Hyperintensities (WMH) are lesions observed in the brain as bright regions in Fluid Attenuated Inversion Recovery (FLAIR) images from Magnetic Resonance Imaging (MRI). Its presence is related to conditions such as aging, small vessel diseases, stroke, depression, and neurodegenerative diseases. Currently, WMH detection is done by specialized radiologists. However, deep learning techniques can learn the patterns from images and later recognize this kind of lesions automatically. This team participated in the MICCAI WMH segmentation challenge, which was released in 2017. A dataset of 60 pairs of human MRI images was provided by the contest, which consisted of T1, FLAIR and ground-truth images per subject. For segmenting the images a 21 layer Convolutional Neural Network-CNN with U-Net architecture was implemented. For validating the model, the contest reserved 110 additional images, which were used to test this method’s accuracy. Results showed an average of 78% accuracy and lesion recall, 74% of lesion f1, 6.24mm of Hausdorff distance, and 28% of absolute percentage difference. In general, the algorithm performance showed promising results, with the validation images not used for training. This work could lead other research teams to push the state of the art in WMH images segmentation.

Paper Nr: 107
Title:

Convolutional Neural Network for Elderly Wandering Prediction in Indoor Scenarios

Authors:

Rafael C. Oliveira, Fabio Barreto and Raphael Abreu

Abstract: This work proposes a way to detect wandering activity of Alzheimer’s patients from path data collected from non-intrusive indoor sensors around the house. Due to the lack of adequate data, we’ve manually generated a dataset of 220 paths using our own developed application. Wandering patterns in the literature are normally identified by visual features (such as loops or random movement), thus our dataset was transformed into images and augmented. Convolutional layers were used on the neural network model since they tend to have good results finding patterns specially on images. The Convolutional Neural Network model was trained with the generated data and achieved an f1 score (relation between precision and recall) of 75%, recall of 60%, and precision of 100% on our 10 sample validation slice.

Short Papers
Paper Nr: 4
Title:

Activity-monitoring in Private Households for Emergency Detection: A Survey of Common Methods and Existing Disaggregable Data Sources

Authors:

Sebastian Wilhelm

Abstract: Ambient-Assisted Living (AAL) technologies can enable the elderly people to live a self-determined life in their own home environment instead of hospitals and retirement homes for a longer period of time. Hence, AAL systems are not only used for everyday support but also for the detection of potential emergency situations and for triggering notification chains. For this purpose the people are usually continuously monitored within their residents by ambient or wearable sensors to detect deviations in their daily behavior. This work surveys common used technologies for Human Activity Recognition (HAR) / Human Presence Detection (HPD), which is the basis for emergency detection. Furthermore, by examining various home automation software, existing data sources from the residential infrastructure, are identified that would be suitable for detecting personal activities.

Paper Nr: 5
Title:

What do You Mean, Doctor? A Knowledge-based Approach for Word Sense Disambiguation of Medical Terminology

Authors:

Erick V. Godinez, Zoltán Szlávik, Edeline Contempré and Robert-Jan Sips

Abstract: Word Sense Disambiguation (WSD) is an essential step for any NLP system; it can improve the performance of a more complex task, like information extraction, named entity linking, among others. Consequently, any error, while disambiguating a term, spreads to later stages with a snowball effect. Knowledge-based strategies for WSD offer the advantage of wider coverage of medical terminology than supervised algorithms. In this research, we present a knowledge-based approach for word sense disambiguation that can use different semantic similarity measures to determine the correct sense of a term in a given context. Our experiments show that when our approach used WordNet-based similarity measures, it achieved a very close performance when using the semantic measures based on word embeddings. We also constructed a small dataset from real-world data, where the feedback received from the annotators made us distinguish between true ambiguous terms and vague terms. This distinction needs to be considered for future research for WSD algorithms and dataset construction. Finally, we analyzed a state-of-the-art dataset with linguistic variables that helped to explain our approach’s performance. Our analysis revealed that texts containing a high score of lexical richness and a high ratio of nouns and adjectives lead to better WSD performance.

Paper Nr: 6
Title:

On the Involvement of Mental Healthcare Professionals in the Co-design of Highly-rated Anxiety Apps

Authors:

Nidal Drissi and Sofia Ouhbi

Abstract: Mobile applications (apps) have the potential to assist people with their mental health issues. They have shown promising results in mitigating many mental health disorders and symptoms, including issues related to anxiety. Mental health apps are based on different approaches, one of which is cognitive behavioral therapy (CBT). However, these solutions still face many concerns and challenges, such as the lack of involvement and inputs of mental healthcare professionals (MHP) in their design and evaluation. This paper focuses on highly-rated CBT-based apps for anxiety and investigates the involvement of MHP in their co-design. Based on the obvious importance of inclusion of mental health professionals in the creation of mental care apps, the following hypothesis was formed: MHP are involved in the design and creation of CBT-based apps for anxiety. To investigate this hypothesis, 23 apps were selected and analysed. Results showed that contrarily to the initial hypothesis, about half of the selected CBT-based apps for anxiety did not involve MHP in their design. Results also showed that the number of installs of the selected apps might be impacted by the involvement of MHP. The average of installs of apps which involved MHP was significantly higher than the average of installs of apps that did not. This might indicate that users tend to trust apps that involve MHP more, which might have impacted their decision to install them. Findings of this study might be of interest to people suffering from anxiety, to help them find apps for anxiety that are based on MHP input, as well as to developers and researchers targeting similar apps.

Paper Nr: 8
Title:

Adoption and Use of Health-related Mobile Applications: A Qualitative Study with Experienced Users

Authors:

Elena Smirnova, Niklas Eriksson and Asle Fagerstrøm

Abstract: Mobile health-related applications (apps) such as physical activity apps and diet apps can help users to implement a more active and healthier lifestyle. This qualitative study investigates experienced users’ triggers to initially download mobile health apps, the drivers that keep them using these types of apps, and the barriers that hinder them from an extended engagement with the apps. Thirteen factors were inductively identified and matched with constructs in theories of technology adoption and use. Also, results from previous studies on mobile health apps were used in the discussion. Life situation, Relevant statistics, and Perceived satisfaction with first health app were identified as initial triggers. Price value, Simplicity, Personalisation, Guidance and Progress based on data, Flexibility, and Social encounters were identified as drivers for continuous use. Perceived risk of personal data, Time-consumption, Limited understanding of health data and Adaption to new routines were identified as barriers for greater engagement with the apps. Managerial implications and further research are also discussed.

Paper Nr: 19
Title:

A Model of Cost and Time-Effective Disease Screening for Non-Communicable Diseases in India

Authors:

Nibras K. Thodika, Srujan Janagam, Smitha T. Kaniyampady, Arkalgud Ramaprasad, Anupama Shetty and Chetan Singai

Abstract: Background: India has the largest burden of the NCDs globally. Screening and identification of clusters with the common risk factors are crucial for early detection, prevention, and control at both individual and population levels. Objective: This study documents a model of cost and time-effective disease screening for NCDs. The model uses a combination of mobile health clinics-based point-of-care diagnostic technologies. The model, for its easy-to-use, cost and time effective operation, should be scalable as a tool for community-based disease screening and population level NCD surveillance. Method: The study documents the materials and processes of NCD screening camps conducted in Bangalore, India. A time and motion study analysis and cost analysis were undertaken to establish the time and cost effectiveness. Results & Discussion: The Study found out a baseline time and cost components for a camp based NCD screening strategy using the mHealth tools and mobile health facilities. This reinforces the potentials of integration of the NCD screening into the public primary health care centres for effective scaling up and achievement of surveillance as well as monitoring and evaluation of the NCD prevention and control programs in the country.

Paper Nr: 21
Title:

Adaptation of Architecture Analyses: An IoT Safety and Security Flaw Assessment Approach

Authors:

Julia Rauscher and Bernhard Bauer

Abstract: Almost everything is connected nowadays or will be in the near future. This trend, called Internet of Things (IoT) or Cyber Physical Systems (CPS), is able to enhance multiple areas e.g. an individual’s life, complex industrial processes or common medical treatments. Though, these improvements frequently affect safety and security critical topics. While this development has many advantages to bring, many challenges arise as well. Most approaches focus on safety and security analyses or monitoring tools determined to apply during run time. These approaches do not consider that plenty of the most dangerous vulnerabilities have to be addressed in the design phase already. Hence, we present an approach to adapt architecture analyses of IoT related areas to provide a holistic tool to assess flaws and possible countermeasures to design a save and secure CPS system.

Paper Nr: 24
Title:

A Machine Learning Approach for Real Time Prediction of Last Minute Medical Appointments No-shows

Authors:

Ricardo Almeida, Nuno A. Silva and André Vasconcelos

Abstract: A no-show is when a patient misses a previously scheduled appointment. No-shows cause an impact in the healthcare sector, decreasing efficiency. When a patient misses an appointment the clinic resource are wasted, postpones his or her chance to get treated for a medical condition and denies medical service to another patient. In this research, machine learning techniques are used to find patterns in healthcare data and make no-show predictions. A no-show prediction model is proposed to integrate machine learning techniques into a model that supports the testing of predictions on different datasets. The model is integrated into an online medical appointment booking platform to allow the models and predictions made, to be saved and integrated into a real-time system. Machine learning techniques are tested using three datasets with different characteristics. Through these tests, the model proposed can find the best features, which are similar in every dataset. The results obtained are compared to other prediction algorithms and techniques.

Paper Nr: 25
Title:

Movement Entropy in a Gardening Design Task as a Diagnostic Marker for Mental Disorders: Results of a Pilot Study

Authors:

Sebastian Unger, Sebastian Appelbaum, Thomas Ostermann and Christina Niedermann

Abstract: Movement, actions, and intentions are important psychological skills in human behavior. Studies have shown correlations between movement activity and a variety of mental disorders. In this context, planning and designing of gardens and outdoor spaces as an intentional activity might play an important role as a marker for mental health. Thus, in this study, 16 subjects (8 female) aged between 19 and 60 were asked to do a gardening task in an experimentally constructed environment while their movement activity was recorded with a camera from a fixed viewpoint. Movement heatmaps and entropy then was calculated and correlated with mental state measured via the Global Severity Index (GSI) of the Brief Symptom Inventory (BSI-18) questionnaire. After finding an optimal grid size of the heatmaps, we were able to find a moderate negative correlation of r = -0.463 between these quantities in an overall of both genders, explaining 21.4 % of variance. After considering the gender of the test group, a noticeable gender effect could be revealed. We found a significant interaction effect of entropy with gender meaning that a lower movement entropy in a gardening task correlates with a higher mental distress for men, but lower for women. Multivariate regression found that this model explained 77.44 % of variance (R = 0.88). Despite of these promising results, further investigations in this area should overcome some limitations in this pilot study in the field of position tracking and movement feature extraction.

Paper Nr: 28
Title:

Homogeneous Ensemble based Support Vector Machine in Breast Cancer Diagnosis

Authors:

Bouchra El Ouassif, Ali Idri and Mohamed Hosni

Abstract: Breast Cancer (BC) is one of the most common forms of cancer and one of the leading causes of mortality among women. Hence, detecting and accurately diagnosing BC at an early stage remain a major factor for women's long-term survival. To this aim, numerous single techniques have been proposed and evaluated for BC classification. However, none of them proved to be suitable in all situations. Currently, ensemble methods have been widely investigated to help diagnosis BC and consists on generating one classification model by combining more than one single technique by means of a combination rule. This paper evaluates homogeneous ensembles whose members are four variants of the Support Vector Machine (SVM) classifier. The four SVM variants used four different kernels: Linear Kernel, Normalized Polynomial Kernel, Radial Basis Function Kernel, and Pearson VII function based Universal Kernel. A Multilayer Perceptron (MLP) classifier is used for combining the outputs of the base classifiers to produce a final decision. Four well-known available BC datasets are used from online repositories. The findings of this study suggest that: (1) ensembles provided a very promising performance compared to its base, and (2) there is no SVM ensemble with a combination of kernels that have better performance in all datasets.

Paper Nr: 31
Title:

I still See You! Inferring Fitness Data from Encrypted Traffic of Wearables

Authors:

Andrei Kazlouski, Thomas Marchioro, Harry Manifavas and Evangelos Markatos

Abstract: In this paper we describe a cyberattack against 2 well-known wearable devices. The attacker presented in this paper is an “honest but curious” Internet Service Provider (ISP) sitting somewhere in the path between the device and the cloud. The ISP launches the attack when the smartbands try to synchronize their data with the permanent cloud storage. By launching its attack, this “honest but curious” ISP is able to learn much personal information about the users of the smartbands, including the frequency of measuring the users’ heart rate and weight; the number and duration of workouts; as well as whether (i) sleep or (ii) steps were recorded on a given day. We show that privacy leaks might occur even when the transferred data are fully encrypted, and the representative mobile application utilizes state-of-the-art security mechanisms: certificate pinning, and source code obfuscation.

Paper Nr: 33
Title:

A 2-minute Fitness Test for Lifestyle Applications: The PhysioFit Task and Its Analysis based on Heart Rate

Authors:

Neide Simões-Capela, Jan Cornelis, Giuseppina Schiavone and Chris Van Hoof

Abstract: Cardio-respiratory fitness (CRF) denotes the health of cardiorespiratory and musculoskeletal systems, thus being important to evaluate effects of (un)healthy lifestyles. Non-exhaustive submaximal fitness tests enable simple, fast, and inexpensive CRF assessment, in situations with low accuracy requirements. An example is the Ruffier-Dickson task (RD), consisting of 30 squats executed within 45 seconds, it estimates a CRF score from heart rate (HR) during the task. Squats, however, are not straightforward for subjects with poor fitness. To overcome this limitation, we developed the PhysioFit task (PF). It entails two minutes of stationary pedaling and employs HR for CRF estimation. PF outcomes were analyzed using RD as benchmark, according to HR changes during the task; CRF scores estimated with methods based on HR; correlation of CRF scores to body composition. The analysis relied on data from 28 subjects who executed both tasks. Although, HR variations during PF were lower relative to RD, PF produced significant changes in HR during pedaling and allowed for significant recovery after one minute. Significant agreement was found between tasks for two CRF scores, and both presented strong negative and positive correlations with fat and muscle percentage, respectively. Preliminary results show that PF is promising towards fast fitness assessments.

Paper Nr: 34
Title:

Morphological Classification of Heartbeats in Compressed ECG

Authors:

Gennaro Laudato, Francesco Picariello, Simone Scalabrino, Ioan Tudosa, Luca De Vito and Rocco Oliveto

Abstract: The number of connected medical devices that are able to acquire, analyze, or transmit health data is continuously increasing. This has allowed the rise of Internet of Medical Things (IoMT). IoMT-systems often need to process a massive amount of data. On the one hand, the colossal amount of data available allows the adoption of machine learning techniques to provide automatic diagnosis. On the other hand, it represents a problem in terms of data storage, data transmission, computational cost, and power consumption. To mitigate such problems, modern IoMT systems are adopting machine learning techniques with compressed sensing methods. Following this line of research, we propose a novel heartbeat morphology classifier, called RENEE, that works on compressed ECG signals. The ECG signal compression is realized by means of 1-bit quantization. We used several machine learning techniques to classify the heartbeats from compressed ECG signals. The obtained results demonstrate that RENEE exhibits comparable results with respect to state-of-the-art methods that achieve the same goal on uncompressed ECG signals.

Paper Nr: 37
Title:

Sociotechnical Challenges of eHealth Technology for Patient Self-management: A Systematic Review

Authors:

Stefan Hochwarter

Abstract: Ageing of society and increase of time spent with chronic conditions challenge the traditional long-term care model. Assistive technology and eHealth are seen to play an important role when addressing these challenges. One prominent example are patient self-management systems. These systems not only transform the way patients with chronic conditions interact with the healthcare system, but also change work practices of care providers. This literature review addresses sociotechnical challenges of eHealth technologies with a strong collaborative component. As a result, four themes are identified and discussed.

Paper Nr: 39
Title:

A Sparse Representation Classification for Noise Robust Wrist-based Fall Detection

Authors:

Farah Othmen, André E. Lazzaretti, Mouna Baklouti, Marwa Jmal and Mohamed Abid

Abstract: Elderly falls are becoming a more crucial and major health problem relatively with the significant growth of the involved population over the years. Wrist-based fall detection solution gained much interest for its comfortable and indoor-outdoor use, yet, a very moving and unstable location to the Inertial measurement unit. Indeed, acquired data might be exposed to random noises challenging the classifier’s reliability to spot falls among other daily activities. In this paper, we address the limits faced by Machine Learning models regarding noisy and overlapped data by proposing a study of the Supervised Dictionary Learning (SDL) technique for on-wrist fall detection. Following the same prior work experimental protocol, the five most popular SDL models were evaluated and compared in performance with two benchmark Machine learning models. The evaluation setup follows two main experiments; processing clean data and casting different additive white Gaussian noise (AWGN). A distinguishable achievement was obtained by the SDL algorithms, of which the Sparse Representation-based Classifier (SRC) algorithm surpass other models especially using noisy data. The latter maintained almost 98% for 0db AWGN versus 96.4% for KNN.

Paper Nr: 41
Title:

Reusability of Interfaces in Healthcare EAI Environments

Authors:

Severin Linecker and Wolfram Wöß

Abstract: Enterprise Application Integration (EAI) and HL7 (Health Level Seven) messaging are well established technologies in healthcare environments. Due to the widely adoption of HL7 messaging, especially the version 2, in the healthcare domain and its flexibility, many vendor specific implementations exist. To integrate these systems, messages have to be adapted to the vendor specific requirements, even if the functionality is nearly the same. This leads to an increasing number of special interfaces and decreased maintainability. This paper shows a generic architecture for reusable interfaces for HL7 messaging by considering reusability at data level and interface level and the results when applied to a real production EAI environment of an austrian healthcare provider.

Paper Nr: 42
Title:

Comfort Evaluation from EEG Dipole Imaging

Authors:

Yuna Shigematsu, Yuta Ueji and Atsushi Ishigame

Abstract: Different people may have different feelings even in the same environment. However, most of the evaluation index in comfort are based on a fixed standard without considering individual differences. In this study, we focus on the preference of comfort, and discussed the dipole imaging of brain waves to evaluate the comfort. The amygdala is said to be one of the parts of the brain related to comfort. In this paper, we stated the relationship between comfort and the area around the amygdala by dipole imaging.

Paper Nr: 43
Title:

Bilingual Emotion Analysis on Social Media throughout the COVID19 Pandemic in Portugal

Authors:

Alina Trifan, Sérgio Matos, Pedro Morgado and José L. Oliveira

Abstract: This paper presents preliminary work on the topic of emotion analysis on Twitter, in the context of the coronavirus pandemic in Portugal. We collected, curated and analyzed covid-related tweets of users in Portugal in order to understand the evolution of the six basic emotions reflected in these tweets. We analyzed tweets written in both English and Portuguese. In this first step of our work we correlate this information with key events of the evolution of the pandemic in Portugal during March, which was the most critical period in Portugal. We do so in an attempt to estimate the online manifestation of the psychological toll that this pandemic has on the overall well-being status of the general population. Our findings show that the sentiment analysis of covid-related tweets is consistent with our hypothesis that negative emotions would intensify as the pandemic progressed. The preliminary results obtained stand as proof of concept that the analysis of real-time tweets or other social media messages through sentiment analysis can be an important tool for behavioural and well-being tracking.

Paper Nr: 45
Title:

Healthcare Provision in the Cloud: An EHR Object Store-based Cloud Used for Emergency

Authors:

Chrysostomos Symvoulidis, Athanasios Kiourtis, Argyro Mavrogiorgou and Dimosthenis Kyriazis

Abstract: EHR Cloud architectures have come a long way within the last years, giving the ability to individuals to store their healthcare data in the cloud, thus being accessible at all times. Though Electronic Health Records (EHR) and Personal Health Records (PHR) sharing technologies have been developed over the last decades, and a lot of the attention is given on the exchange of healthcare data between organizations and healthcare institutions, less emphasis has been put in the services regarding the exchange of such data between individuals and healthcare professionals and the issues that this gap creates are yet to be answered. To this end, in this paper, we introduce an EHR Cloud-based system that utilizes an Object Storage architecture to store healthcare data, and provides the ability to authenticated healthcare professionals to gain access if needed during an emergency, in an automated yet secure way, for accelerated health services provision. The proposed approach is evaluated and the results are presented in order to justify the rationale behind its design.

Paper Nr: 48
Title:

Physical Burden in Manual Patient Handling: Quantification of Lower Limb EMG Muscle Activation Patterns of Healthy Individuals Lifting Different Loads Ergonomically

Authors:

Anna Brinkmann, Conrad F. Böhlen, Sandra Hellmers, Ole Meyer, Rebecca Diekmann and Andreas Hein

Abstract: Manual patient handling is a challenging part of daily care and leads to high mechanical loads as well as to the development of degenerative diseases, e.g. lower back pain. To prevent musculoskeletal overload effects, the use of ergonomic working techniques is essential as well as improving caregivers’ functional ability. However, most of the studies do not consider these aspects and biomechanical evaluations including dynamic electromyography (EMG) are rarely analyzed. In this work, we focus on the quantification of lower limb EMG muscle activation patterns of healthy caregiver students in an experimental setup. The extent of lifting different loads ergonomically is analyzed and similarities/dissimilarities of dynamic EMG data of three lower limb muscles are investigated via cross-correlation calculation. One of the main findings of our investigation is an indication of a more consistent mean activity of the quadriceps and hamstring musculature, as the load to be lifted increases. Furthermore, we found an intra-as well as an interindividual similarity of EMG muscle activation patterns regarding time and shape of the signals generated during all of the conducted lifting tasks with a predominantly high cross-correlation coefficient for the selected muscles of the lower limb.

Paper Nr: 50
Title:

Assessing COVID-19 Impacts on College Students via Automated Processing of Free-form Text

Authors:

Ravi Sharma, Sri D. Pagadala, Pratool Bharti, Sriram Chellappan, Trine Schmidt and Raj Goyal

Abstract: In this paper, we report experimental results on assessing the impact of COVID-19 on college students by processing free-form texts generated by them. By free-form texts, we mean textual entries posted by college students (enrolled in a four year US college) via an app specifically designed to assess and improve their mental health. Using a dataset comprising of more than 9000 textual entries from 1451 students collected over four months (split between pre and post COVID-19), and established NLP techniques, a) we assess how topics of most interest to student change between pre and post COVID-19, and b) we assess the sentiments that students exhibit in each topic between pre and post COVID-19. Our analysis reveals that topics like Education became noticeably less important to students post COVID-19, while Health became much more trending. We also found that across all topics, negative sentiment among students post COVID-19 was much higher compared to pre-COVID-19. We expect our study to have an impact on policy-makers in higher education across several spectra, including college administrators, teachers, parents, and mental health counselors.

Paper Nr: 51
Title:

Custom FHIR Resources Definition of Detailed Radiation Information for Dose Management Systems

Authors:

Abderrazek Boufahja, Steven Nichols and Vincent Pangon

Abstract: Medical diagnostic imaging dose management systems aggregate and calculate irradiation dose generated by acquisition modalities, collected through standardized methods such as DICOM®, HL7® or proprietary interfaces. Irradiation dose information is valuable to multiple stakeholders, such as, general practitioners (GP), nationalized dose registries, patient facing applications, and information systems, such as, the Radiology Information System (RIS) or Electronic Health Record (EHR). For Medical Physicists, the radiation data is used to perform patient cohort and statistical analysis as part of a dose management program. However, there is no standardized, lightweight method to exchange the collected dose information with third party applications, through RESTful APIs. In this paper, we define a methodology to expose the content of the Radiation Dose DICOM® SR data models as custom HL7® FHIR® resources. This methodology leverages the strength of FHIR® in defining and exchanging resources, and the strength of the DICOM® SR data models, as their structure is implemented, maintained, and tested by dozens of modality providers.}

Paper Nr: 53
Title:

Usability Assessment of an Intraoperative Planning Software

Authors:

Federico Sternini, Giuseppe Isu, Giada Iannizzi, Diego Manfrin, Noemi Stuppia, Federica Rusinà and Alice Ravizza

Abstract: Usability is a crucial aspect of medical device safety. The brand-new European Regulation requires the manufacturer to assess the usability of the new medical devices. In this study, we evaluate the usability of a new medical device intended to assist the intraoperative planning with the visualization of 3d patient-specific organ models. The usability study started from the early stage of the device design and iterated through an early formative, completed with desk-based activities, late formative, completed with a focus group, and summative phase, that comprised a user test, and questionnaire filling. The identified usability issues are mitigated, the safety of the device user interface is confirmed and the training contents are defined and confirmed. Additional information regarding the user experience is collected and analyzed to identify further improvements of the device.

Paper Nr: 54
Title:

Needs, Functions, and Technologies of Technical Assistance Systems in Nursing Context: A Systematic Review

Authors:

Alexander Hammer, Bastian Wollschlaeger, Martin Schmidt and Lena Otto

Abstract: Due to the ageing of society, health insurers and care sectors in many Western countries are facing major challenges. Technical Assistance Systems (TAS) could have the potential to ease the situation, while at the same time promoting the independence and self-determination of care-dependent people. However, TAS have not yet been fully established in nursing. Reasons for this include an inadequate systematisation of the research and development area and the lack of uniform terminology, which leads to poor comparability and thus to missing financing models. To tackle this condition and help to select functions and technologies based on needs, we conduct a systematic review identifying needs, functions, and technologies as areas of interest addressed in 50 evaluated TAS approaches. Further, this work assesses gaps in TAS research and aims to create a uniform understanding of assistance functions.

Paper Nr: 57
Title:

Automated Medical Reporting: From Multimodal Inputs to Medical Reports through Knowledge Graphs

Authors:

Lientje Maas, Adriaan Kisjes, Iman Hashemi, Floris Heijmans, Fabiano Dalpiaz, Sandra Van Dulmen and Sjaak Brinkkemper

Abstract: Care providers generally experience a high workload mainly due to the large amount of time required for adequate documentation. This paper presents our visionary idea of real-time automated medical reporting through the integration of speech and action recognition technology with knowledge-based summarization of the interaction between care provider and patient. We introduce the Patient Medical Graph as a formal representation of the dialogue and actions during a medical consultation. This knowledge graph represents human anatomical entities, symptoms, medical observations, diagnoses and treatment plans. The formal representation enables automated preparation of a consultation report by means of sentence plans to generate natural language. The architecture and functionality of the Care2Report prototype illustrate our vision of automated reporting of human communication and activities using knowledge graphs and NLP tools.

Paper Nr: 61
Title:

Investigating the Impediments to Accessing Reliable, Timely and Integrated Electronic Patient Data in Healthcare Sites in Uganda

Authors:

Andrew E. Alunyu, Joseph Wamema, Achilles Kiwanuka, Bagyendera Moses, Mercy Amiyo, Andrew Kambugu and Josephine Nabukenya

Abstract: The purpose of collecting patient data is to support their care and wellbeing. Patient-centred care is attained by securely availing all records about the patient whenever it's necessary to the right persons and at the right time. However, healthcare providers have often failed to share integrated patient data on time due to limitations in accessing reliable patient data required to inform care/treatment decisions. This study aimed to investigate impediments to accessing reliable, timely and integrated patient data through investigating the processes for collection, analysis, and presentation of data across various healthcare sites in Uganda. A cross-sectional study design was followed, and data was collected from purposively selected National level (policymakers) and Sub-national level (health facilities). The field findings indicate various impediments to accessing patient data including but not limited to inadequate mechanisms for electronic health data collection, storage and access, non-standardised health data sharing mechanisms, inadequate Health Information System (HIS) and Information and Communication Technology (ICT) infrastructure, and inadequate skills, knowledge and training. Other impediments included; insufficient security and privacy measures, weak eHealth governance, and inadequate management support. Accordingly, these have negatively impacted on patient data use and quality of patient care in Uganda.

Paper Nr: 65
Title:

Ambulatory Assessment of Mental Health and Well-being using an Experience Sampling Methodology: Pipeline

Authors:

Jaymar Soriano, Alyanna Gacutan, Martina S. Casiano, Jeanne N. Magpantay, Allure D. Tanquintic, Gina R. Tongco-Rosario, Grazianne-Geneve Mendoza and Christie Sio

Abstract: Mental health disorders are prevalent in our society today as they affect one in four people all over the world, according to the World Health Organization. This necessitates a proactive method of mental health assessment. Clinical assessments and paper-and-pen reporting are usually done through retrospective reports, which are subject to memory bias. With the advancement in technology and smartphones becoming an inherent and integral part of day-to-day life, ambulatory assessment of mental health and well-being would be greatly improved. In this study, we analyze different methods of mental health assessment, their respective benefits and drawbacks, and from which we propose a pipeline based on Experience Sampling Methodology (ESM). The pipeline is composed of a web application for therapists and a mobile application for patients. The therapist creates ESM-based assessments to their patients using the web application that communicates with the mobile application through an application programming interface. This pipeline aims to overcome retrospective biases in assessing the patient’s mental health and well-being by using more reliable behavioural patterns from the data. Sophisticated data encryption may be utilized to ensure patient-therapist confidentiality. The same system is also designed to be used by psychologists to send ESM-based surveys to their intended participants and perform statistical analysis from the respondents’ data, allowing improved data security for the respondents. With this capability, generation of data would be faster and safer, and more research can be done to improve and accelerate analysis and diagnosis of mental health and well-being.

Paper Nr: 70
Title:

Toward a Compare and Contrast Framework for COVID-19 Contact Tracing Mobile Applications: A Look at Usability

Authors:

Cristiano Storni, Damyanka Tsvyatkova, Ita Richardson, Jim Buckley, Manzar Abbas, Sarah Beecham, Muslim Chochlov, Brian Fitzgerald, Liam Glynn, Kevin Johnson, John Laffey, Bairbre McNicholas, Bashar Nuseibeh, James O’Connell, Derek O’Keeffe, Ian R. O'Keeffe, Mike O’Callaghan, Abdul Razzaq, Kaavya Rekanar, Andrew Simpkin, Jane Walsh and Thomas Welsh

Abstract: This paper reports on the progress in the project COVIGILANT, which is aimed at developing an evaluation taxonomy for Contact Tracing Applications (CTAs) for COVID-19. Specifically, this article describes the development of Usability, one pillar of the COVIGILANT taxonomy, discussing the classification and decision-making processes, and the initial model validation. The validation process was undertaken in two stages. First, we validated how the Usability pillar could be used to evaluate the Irish Health Services Executive (HSE) COVID-19 CTA. While this supported many of the attributes that we had within the Usability pillar, it also identified issues. We made amendments based on these, and undertook a second study, this time evaluating 4 CTAs used in other countries. This has led to the completion of the Usability pillar, which can now be used to evaluate global CTAs.

Paper Nr: 71
Title:

I-ODA, Real-world Multi-modal Longitudinal Data for Ophthalmic Applications

Authors:

Nooshin Mojab, Vahid Noroozi, Abdullah Aleem, Manoj P. Nallabothula, Joseph Baker, Dimitri T. Azar, Mark Rosenblatt, R. P. Chan, Darvin Yi, Philip S. Yu and Joelle A. Hallak

Abstract: Data from clinical real-world settings is characterized by variability in quality, machine-type, setting, and source. One of the primary goals of medical computer vision is to develop and validate artificial intelligence (AI) based algorithms on real-world data enabling clinical translations. However, despite the exponential growth in AI based applications in healthcare, specifically in ophthalmology, translations to clinical settings remain challenging. Limited access to adequate and diverse real-world data inhibits the development and validation of translatable algorithms. In this paper, we present a new multi-modal longitudinal ophthalmic imaging dataset, the Illinois Ophthalmic Database Atlas (I-ODA), with the goal of advancing state-of-the-art computer vision applications in ophthalmology, and improving upon the translatable capacity of AI based applications across different clinical settings. We present the infrastructure employed to collect, annotate, and anonymize images from multiple sources, demonstrating the complexity of real-world retrospective data and its limitations. I-ODA includes 12 imaging modalities with a total of 3,668,649 ophthalmic images of 33,876 individuals from the Department of Ophthalmology and Visual Sciences at the Illinois Eye and Ear Infirmary of the University of Illinois Chicago (UIC) over the course of 12 years.

Paper Nr: 72
Title:

Efficacy of Augmented Reality-based Virtual Hiking in Cardiorespiratory Endurance: A Pilot Study

Authors:

Muhammad A. Ahmad, Honorato Sousa, Élvio R. Quintal and Sergi Bermúdez i Badia

Abstract: Exergames can be used to overcome a sedentary lifestyle. Virtual Reality (VR) has made exergames successful, and they can be used to increase heart rate, but some limitations are found, such as the adaptation of the heart rate in exergames to the player's fitness profile. VR technology has been used to simulate virtual cycling and walking experiences. We designed and developed an exergame' Virtual Levadas' in a cave-based VR environment to simulate the Levadas hiking tracks. They are the main attraction for tourists in Madeira Island, Portugal. This study's main objective was to assess player exertion, usability, participation, and realism of the simulation of the Levadas tracks. We performed this study with 13 participants who played Virtual Levadas for 6 minutes and found a significant increase in player's average physical activity and heart rate. Overall, our results demonstrate that Virtual Levada's exergame provides a higher exertion level, immersion, and realism of the virtual environment than the literature.

Paper Nr: 73
Title:

SeVA: An AI Solution for Age Friendly Care of Hospitalized Older Adults

Authors:

Chongke Wu, Jeno Szep, Salim Hariri, Nimit K. Agarwal, Sumit K. Agarwal and Carlos Nevarez

Abstract: As a dangerous syndrome, delirium affects more than 50% of hospitalized older adults and has an economic burden of 164 billion US dollars per year. It is crucial to prevent, identify and treat this syndrome systematically on all hospitalized patients to prevent its short and long-term complications. Currently, there are no AI-based tools being utilized at a large scale focused on delirium management in hospital settings. The advancement of the Internet of Things in the medical arena can be leveraged to help clinical teams managing the care of patients in the hospital. The renaissance of Artificial Intelligence brings the chance to analyze a large amount of monitoring data. Deep neural networks like Convolutional Neural Network and Recurrent Neural Network revolutionize the fields of Computer Vision and Natural Language Processing. Deep learning tasks like action recognition and language understanding can be incorporated into the routine workflow of healthcare staff to improve care. By leveraging AI and deep learning techniques, we have developed a chatbot based monitoring system (that we refer to as SeVA) to improve the workload of the medical staff by using an Artificial Emotional Intelligence platform. The SeVA platform includes two mobile applications that provide timely patient monitoring, regular nursing checks, and health status recording features. We demonstrate the current progress of deploying the SeVA platform in a healthcare setting.

Paper Nr: 74
Title:

De-identification of Clinical Text for Secondary Use: Research Issues

Authors:

Hanna Berg, Aron Henriksson, Uno Fors and Hercules Dalianis

Abstract: Privacy is challenged by both advances in AI-related technologies and recently introduced legal regulations. The problem of privacy has been extensively studied within the privacy community, but has largely focused on methods for protecting and assessing the privacy of structured data. Research aiming to protect the integrity of patients based on clinical text has primarily referred to US law and relied on automatically recognising predetermined, both direct and indirect, identifiers. This article discusses the various challenges concerning the re-use of unstructured clinical data, in particular in the form of clinical text, and focuses on ambiguous and vague terminology, how different legislation affects the requirements for de-identification, differences between methods for unstructured and structured data, the impact of approaches based on named entity recognition and replacing sensitive data with surrogates, as well as the lack of measures for usability and re-identification risk.

Paper Nr: 75
Title:

A near Complete Adoption of Electronic Health Records System in the U.S. Lacks Interoperability and Physician Satisfaction

Authors:

Raghid El-Yafouri and Leslie Klieb

Abstract: In this position paper, the satisfaction of health care providers with electronic health record (EHR) systems is discussed. Based on a survey just before the deadline for EHR adoption incentive by the U.S. government, we conclude that too many physicians and medical care providers do not like the current state of the systems, feel forced to use them, and get insufficient benefits from them. We urge all involved parties to collaborate and design systems that are more in agreement with the practices of the different parties and specialties in the health care industry.

Paper Nr: 76
Title:

Handling Comparison between a Human and a Patient Simulator for Nursing Care Related Physical Human-robot Interaction

Authors:

Christian Kowalski, Pascal Gliesche, Conrad F. Böhlen, Anna Brinkmann and Andreas Hein

Abstract: The occurrence of musculoskeletal diseases among nursing staff leads to an early withdrawal from the profession, which reinforces the already existing lack of caregivers. To counteract this problem, we would like to provide physical relief through robotic assistance at the bedside. However, the problem arises that for safety reasons robotic assistance concepts should not be tested on humans at first. In this case, patient simulators with an average person’s weight can function as a substitute. For the best results, both cases should behave very similarly when evaluating robot assistance concepts so that the transfer from patient simulator to human is small and therefore no major adjustments need to be made. To measure this potential difference, we have compared the handling properties of both cases in this paper. We examined force measuring platform data while a nurse mobilized an 80 kg human and a patient simulator from the back to the side. The experimental results show that moving a patient simulator is more physically demanding compared to moving a human with similar weight and that conventional collaborative lightweight robots are able to push and move a patient simulator that is weighing far higher than the robot’s actual payload suggests.

Paper Nr: 78
Title:

bwHealthApp: A Software System to Support Personalized Medicine by Individual Monitoring of Vital Parameters of Outpatients

Authors:

Philip Storz, Sandra Wickner, Benjamin Batt, Johannes Schuh, Denise Junger, Yvonne Möller, Nisar Malek and Christian Thies

Abstract: Continuous monitoring of individual vital parameters can provide information for the assessment of one’s health and indications of medical problems in the context of personalized medicine. Correlations between parameters and health issues are to be evaluated. As one project in this topic area, a telemedicine platform is implemented to gather data of outpatients via wearables and accumulate them for physicians and researchers to review. This work extracts requirements, draws use case scenarios, and shows the current system architecture consisting of a patient application, a physician application with a web server, and a backend server application. In further work, the prototype will assist to develop a vendor-free and open monitoring solution. A conclusion on functionality and usability will be evaluated in an imminent first study.

Paper Nr: 80
Title:

Online Decision Support Tool that Explains Temporal Prediction of Activities of Daily Living (ADL)

Authors:

Janusz Wojtusiak, Negin Asadzaehzanjani, Cari Levy, Farrokh Alemi and Allison E. Williams

Abstract: This paper presents an online decision support tool that can be used to assess and predict functional abilities in terms of nine Activities of Daily Living (ADLs) up to one year ahead. The tool is based on previously developed Computational Barthel Index (CBIT) and has been rebuilt using Gradient Boost (GB) models with average Area under ROC (AUC) of 0.79 (0.77-0.80), accuracy of 0.74 (0.70-0.79), recall of 0.78 (0.58-0.93), and precision of 0.75 (0.67-0.82) when evaluating ADLs for new patients. When re-evaluating patients, the models achieved AUC 0.95 (0.94-0.96), accuracy of 0.91 (0.90-0.92), recall of 0.91 (0.86-0.95), and precision of 0.92 (0.88-0.94). The decision support tool has been equipped with a prediction explanation module that calculates and visualizes influence of patient characteristics on the predicted values. The explanation approach focuses on patient characteristics present in the data, rather than all attributes used to construct models. The tool has been implemented in Python programming language using Flask Web framework and is accessible through a website or an Application Programming Interface (API).

Paper Nr: 81
Title:

The Need for Medical Professionals to Join Patients in the Online Health Social Media Discourse

Authors:

Hamman Samuel, Fahim Hassan and Osmar Zaíane

Abstract: Health social media is frequently used by e-patients for seeking health information online for self-diagnosis, self-treatment, and self-education. Health social media also provides various benefits for patients and laypersons, such as allowing users to be part of virtual support groups, having quick access to advice, and the convenience of access via the Internet. At the same time, it raises concerns about misinformation being propagated by laypersons without professional medical expertise, especially during pandemics like COVID-19, leading to an infodemic. There are only a handful of health social media websites that allow medical professionals to participate in discussions with patients online. We postulate that the modern face of medicine and healthcare needs medical professionals to be included in the online patient discourse so misinformation can be addressed early on and head on. To this end, we propose a new and free health social network named Cardea that is under development which aims to bring patients, laypersons, and medical professionals together on the world wide web. Users can share experiences, ask questions, and get answers in three streamlined environments: Patient to Patient (P2P), Patient to Medic (P2M), and Medic to Medic (M2M). While there are several forums that cater specifically to patient-patient discussions or medic-medic connections, Cardea’s added value is in providing a unified portal for both patients and medics, as well as enabling interactions between patients and medics. Moreover, Cardea applies machine learning, information retrieval, and natural language processing methods to promote credible health information and demote misinformation. At the same time, with enhanced veracity, anonymity, and privacy controls, the vision of Cardea is for e-patients to confidently share experiences and opinions without being stigmatized or compromising their right to privacy. Our hope is to generate discussion and gather insights from other researchers on developing Cardea.

Paper Nr: 82
Title:

Human Activity Recognition using Deep Learning Models on Smartphones and Smartwatches Sensor Data

Authors:

Bolu Oluwalade, Sunil Neela, Judy Wawira, Tobiloba Adejumo and Saptarshi Purkayastha

Abstract: In recent years, human activity recognition has garnered considerable attention both in industrial and academic research because of the wide deployment of sensors, such as accelerometers and gyroscopes, in products such as smartphones and smartwatches. Activity recognition is currently applied in various fields where valuable information about an individual’s functional ability and lifestyle is needed. In this study, we used the popular WISDM dataset for activity recognition. Using multivariate analysis of covariance (MANCOVA), we established a statistically significant difference (p < 0.05) between the data generated from the sensors embedded in smartphones and smartwatches. By doing this, we show that smartphones and smartwatches don’t capture data in the same way due to the location where they are worn. We deployed several neural network architectures to classify 15 different hand and non-hand oriented activities. These models include Long short-term memory (LSTM), Bi-directional Long short-term memory (BiLSTM), Convolutional Neural Network (CNN), and Convolutional LSTM (ConvLSTM). The developed models performed best with watch accelerometer data. Also, we saw that the classification precision obtained with the convolutional input classifiers (CNN and ConvLSTM) was higher than the end-to-end LSTM classifier in 12 of the 15 activities. Additionally, the CNN model for the watch accelerometer was better able to classify non-hand oriented activities when compared to hand-oriented activities.

Paper Nr: 85
Title:

COVIDDX: AI-based Clinical Decision Support System for Learning COVID-19 Disease Representations from Multimodal Patient Data

Authors:

Veena Mayya, Karthik K., Sowmya S. Kamath, Krishnananda Karadka and Jayakumar Jeganathan

Abstract: The COVID-19 pandemic has affected the world on a global scale, infecting nearly 68 million people across the world, with over 1.5 million fatalities as of December 2020. A cost-effective early-screening strategy is crucial to prevent new outbreaks and to curtail the rapid spread. Chest X-ray images have been widely used to diagnose various lung conditions such as pneumonia, emphysema, broken ribs and cancer. In this work, we explore the utility of chest X-ray images and available expert-written diagnosis reports, for training neural network models to learn disease representations for diagnosis of COVID-19. A manually curated dataset consisting of 450 chest X-rays of COVID-19 patients and 2,000 non-COVID cases, along with their diagnosis reports were collected from reputed online sources. Convolutional neural network models were trained on this multimodal dataset, for prediction of COVID-19 induced pneumonia. A comprehensive clinical decision support system powered by ensemble deep learning models (CADNN) is designed and deployed on the weba. The system also provides a relevance feedback mechanism through which it learns multimodal COVID-19 representations for supporting clinical decisions.

Paper Nr: 91
Title:

Reproducibility, Transparency and Evaluation of Machine Learning in Health Applications

Authors:

Janusz Wojtusiak

Abstract: This paper argues for the importance of detailed reporting of results of machine learning modeling applied in medical, healthcare and health applications. It describes ten criteria under which results of modeling should be reported. The ten proposed criteria are experimental design, statistical model evaluation, model calibration, top predictors, global sensitivity analysis, decision curve analysis, global model explanation, local prediction explanation, programming interface and source code. The criteria are discussed and illustrated in the context of existing models. The goal of the reporting is to ensure that results are reproducible, and models gain trust of end users. A brief checklist is provided to help facilitate model evaluation.

Paper Nr: 93
Title:

Supporting Childbirth Knowledge Acquisition and Decision-making through Digital Communication Technology: The Research Design of an Ongoing Study following a Mixed-Method Approach

Authors:

Carla V. Leite and Ana M. Almeida

Abstract: This paper describes the research design of an ongoing study that overlaps three main fields: technology, health, and social science. This transdisciplinarity approach naturally brings challenges to the methodological plan, which this paper presents, and aims to guide the creation, validation and evaluation of a digital decision aid, and its comparison to a paper-based solution. Through the data collection from different natures, it is expected to be possible to understand the different sources, channels and formats of content that can contribute for childbirth knowledge acquisition; if communication can be facilitated between expectant parents, health care professionals, and childbirth educators; and ultimately, if the tool could provide a mean to create a document regarding birth preferences.

Paper Nr: 97
Title:

Exploring Media Portrayals of People with Mental Disorders using NLP

Authors:

Swapna Gottipati, Mark Chong, Andrew W. Kiat and Benny H. Kawidiredjo

Abstract: Media plays an important role in creating an impact in society. Several studies show that news media and entertainment channels, at times may create overwhelming images of the mental illness that emphasize criminality and dangerousness. The consequences of such negative impact may impact the audience with stigma and on the other hand, they impair the self-esteem and help-seeking behavior of the people with mental disorders. This is the first study to examine the Singapore media’s portrayal of persons with mental disorders (MDs) using text analytics and natural language processing. To date, most studies on media portrayal of people with MDs have been conducted in developed Western countries. This study found that media articles on MDs in Singapore were largely negative in sentiment; even quotes from experts contain aspects of stigma. In addition, crime-related articles on MDs accounted for a significant portion of the corpus. Our model is also extended to detect positive health articles that discuss recovery and motivation. We further developed a stigma classifier based on the machine learning algorithms and text mining techniques. The classifier based on the XGBoosts performed best with an F1-score around 76%.

Paper Nr: 99
Title:

In Search of a Conversational User Interface for Personal Health Assistance

Authors:

Mathias W. Jesse, Claudia Steinberger and Peter Schartner

Abstract: Conversational user interfaces (CUI) cause a paradigm shift in the interaction between user and machine. The machine is operated via structured dialogues or partly or entirely via human language. Voice assistants that understand and process spoken natural language are increasingly being used. The currently available conversational technologies (CTs) for voice assistants range from well-known commercial technologies to quite well-known open source platforms. The suitability of a CT for a particular application domain depends heavily on its specific requirements. In this paper, we focus on the selection of CTs for the development of CUIs for the elderly to assist them in their health management. We (1) propose criteria for CT selection in the domain of personal health management for the elderly, (2) analyze commercial and open source representatives according to these criteria and (3) we evaluate the most suitable candidates for CUI development.

Paper Nr: 100
Title:

A Survey of Possibilities and Challenges with AR/VR/MR and Gamification Usage in Healthcare

Authors:

Yu Fu, Yan Hu, Veronica Sundstedt and Cecilia Fagerström

Abstract: Software and applications of augmented reality (AR), virtual reality (VR), and mixed reality (MR) technology combined with game/gamification techniques in healthcare have increasingly been studied in academia. However, there is a need to explore the usage, challenges and opportunities of AR/VR/MR game/gamification software/applications in the healthcare system. To explore this, we present an online survey conducted in the healthcare-relevant system (including hospital-based system, homecare-based system, institute and university, and industry). Based on the answers, we found examples of digital games and AR/VR/MR applications used in healthcare, as well as some general information (name and feature, purpose, target user, and use occasion), usage situation, and user experience. This presented survey is beneficial for both researchers and developers in computer science and medical science. It can familiarise them with existing products and their current use, advantages and potential issues of AR/VR/MR and game applications in healthcare. In future work, the survey would be extended to obtain other user experiences and feedback of AR/VR/MR techniques and game/gamification technology applied to healthcare, as well as to study how to overcome the challenges, and develop the opportunities further.

Paper Nr: 101
Title:

Multi-resistant Bacterial Infection Surveillance using a Graph Database with Spatio-temporal Information

Authors:

Lorena Pujante, Manuel Campos, Jose M. Juarez, Bernardo Canovas-Segura and Antonio Morales

Abstract: Some of epidemiologists’ efforts in dealing with multi-resistant bacterial infections acquired in healthcare settings focus on tracing patient’s activities, carrying out a contact analysis, and identifying the main risk factors that lead the appearance of these infections. Most contact analysis studies assume information is stored in conventional relational databases. To date, little attention has been paid to other storage paradigms. This paper explores the potential of graph databases to establish the complex relations required to compute contact analysis. We discuss the advances in modelling of the temporal and the spatial information of the Electronic Health Record that has to be introduced in the graph database. In this position paper we propose three points for discussion: advances in formal modelling, specific algorithms for graph analysis, and visualisation tools.

Paper Nr: 102
Title:

Towards Semantic Integration for Explainable Artificial Intelligence in the Biomedical Domain

Authors:

Catia Pesquita

Abstract: Explainable artificial intelligence typically focuses on data-based explanations, lacking the semantic context needed to produce human-centric explanations. This is especially relevant in healthcare and life sciences where the heterogeneity in both data sources and user expertise, and the underlying complexity of the domain and applications poses serious challenges. The Semantic Web represents an unparalleled opportunity in this area: it provides large amounts of freely available data in the form of Knowledge Graphs, which link data to ontologies, and can thus act as background knowledge for building explanations closer to human conceptualizations. In particular, knowledge graphs support the computation of semantic similarity between objects, providing an understanding of why certain objects are considered similar or different. This is a basic aspect of explainability and is at the core of many machine learning applications. However, when data covers multiple domains, it may be necessary to integrate different ontologies to cover the full semantic landscape of the underlying data. We propose a methodology for semantic explanations in the biomedical domain that is based on the semantic annotation and integration of heterogenous data into a common semantic landscape that supports semantic similarity assessments. This methodology builds upon state of the art semantic web technologies and produces post-hoc explanations that are independent of the machine learning method employed.

Paper Nr: 105
Title:

Towards Customized Medicine with Open-source Applications in Developing Countries: Foot Drop and Transtibial Prosthesis

Authors:

Livingston C. Valladares, Juan Lamán, Xavier Riccio, David Aucancela, Francis R. Loayza and Gilbert Sotomayor

Abstract: Alterations in the normal gait can be enhanced to improve patients’ quality of life. Although several devices improve these conditions, the technology to diagnose and create solutions is expensive. The present work focuses on developing a methodology to use free software and hardware to create solutions. The process starts gathering and analyzing the patient’s clinical data; then analyze the human motion kinematics of the patient, so it is possible to customize and manufacture either an orthotic or prosthetic device. With the aim of implementing the methodology, two cases of study are presented in this work. The patient with foot drop presented an angular difference between the ankle and the toe of 10.10◦ ± 4.76◦ , which was corrected throughout the spring-like behavior of the material used for the 3D printing process. Further, the prosthetic device was a design with an ankle joint that allows the plantarflexion and dorsiflexion angles of 30° and 25°, respectively. Therefore, this methodology allows the diagnosing of the angular difference between joints during the normal gait and how to create either orthotic or prosthetic devices to reduce them. Hence, the present work aims to open doors towards the customization of medicine and rehabilitation, especially in developing countries.

Paper Nr: 106
Title:

Simple Matrix Factorization Collaborative Filtering for Drug Repositioning on Cell Lines

Authors:

Iván Carrera, Eduardo Tejera and Inês Dutra

Abstract: The discovery of new biological interactions, such as interactions between drugs and cell lines, can improve the way drugs are developed. Recently, there has been important interest for predicting interactions between drugs and targets using recommender systems; and more specifically, using recommender systems to predict drug activity on cellular lines. In this work, we present a simple and straightforward approach for the discovery of interactions between drugs and cellular lines using collaborative filtering. We represent cellular lines by their drug affinity profile, and correspondingly, represent drugs by their cell line affinity profile in a single interaction matrix. Using simple matrix factorization, we predicted previously unknown values, minimizing the regularized squared error. We build a comprehensive dataset with information from the ChEMBL database. Our dataset comprises 300,000+ molecules, 1,200+ cellular lines, and 3,000,000+ reported activities. We have been able to successfully predict drug activity, and evaluate the performance of our model via utility, achieving an Area Under ROC Curve (AUROC) of near 0.9.

Paper Nr: 10
Title:

Predictive Clustering Learning Algorithms for Stroke Patients Discharge Planning

Authors:

Luigi Lella, Luana Gentile, Christian Pristipino and Danilo Toni

Abstract: Stroke patients discharge planning is a complex task that could be carried out by the use of a suitable decision support system. Such a platform should be based on unsupervised machine learning algorithms to reach the best results. More specifically, in this kind of prediction task clustering learning algorithms seem to perform better than the other unsupervised models. These algorithms are able to independently subdivide the treated clinical cases into groups, and they can serve to discover interesting correlations among the clinical variables taken into account and to improve the prediction accuracy of the treatment outcome. This work aims to compare the prediction accuracy of a particular clustering learning algorithm, the Growing Neural Gas, with the prediction accuracy of other supervised and unsupervised algorithms used in stroke patients discharge planning. This machine learning model is also able to accurately identify the input space topology. In other words it is characterized by the ability to independently select a subset of attributes to be taken into consideration in order to correctly perform any predictive task.

Paper Nr: 12
Title:

e-Health Implementation in Lebanese Teaching Hospitals: What Can We Learn from Their Successes and Imminent Challenges

Authors:

Nabil G. Badr, Elissar Chami and Michele K. Asmar

Abstract: We set on an exploration to learn from two major implementations of eHealth in teaching hospitals in Lebanon. After an explorative qualitative empirical work; we summarize learnings from these successes and present them as a backdrop for future studies. The main value of this study is in the discovery of the importance of technical and process readiness with an emphasis on dedicated stakeholder engagement to guarantee the successful outcome. The focus on the patient journey and the wellbeing of the practitioner in the new digital ecosystem are as important as financial preparation and infrastructure readiness. Researchers are still pondering the unintended consequences of EMR implementations and the effect of such implementations on the stakeholders of the healthcare ecosystem. In this paper, we argue that the road to success in the implementation of eHealth must be through stakeholder engagement as a means to increase the satisfaction of practitioners with the new work environment.

Paper Nr: 26
Title:

Responding to COVID-19: Potential Hospital-at-Home Solutions to Re-configure the Healthcare Service Ecosystem

Authors:

Nabil G. Badr, Luca Carrubbo and Marguerita Ruberto

Abstract: An effective Healthcare Service Ecosystem must emphasize the notion of well-being co-creation which entails a dynamic interplay of actors, in face of the challenges, with their ability to use the available resource pools, at the different system levels. An appropriate response, largely avoiding any crisis, depends on a society's resilience and the related response of actors in the reconfiguration of resources. Originally considered luxury and for the fortunate few who could afford the learning curve, Hospitalization-at-Home (HaH) recently approached a new normal with a positive impact to health outcomes. Nowadays, hospitals have had to reconfigure their health services to reduce the workload of caregivers during the COVID-19 outbreak. Our use case can be a lesson for the adaptation of technology for patient empowerment allowing patients to interact with their care ecosystem while at their home.

Paper Nr: 29
Title:

Player-Type-based Personalization of Gamification in Fitness Apps

Authors:

Nadine Sienel, Patrick Münster and Gottfried Zimmermann

Abstract: This paper examines the effect of personalized gamification on an individual’s motivation in the context of fitness apps. In a first study, we evaluate the four categorization models "Bartle Player Types", "Big Five", "Hexad User Types", and "BrainHex" on their ability to predict individual gamification preferences of users and develop a new prediction model called “MoMo”. Bartle, BrainHex, and MoMo are validated empirically in a second study, employing off-the-shelf fitness apps with gamification elements. The results of both studies indicate that a prediction is possible using the categorization models. Among all models, MoMo performs best in predicting individual gamification preferences, followed by BrainHex. Results of the second study indicate that, although the models MoMo and BrainHex perform better in predicting the theoretical rating of gamification elements than the random model, the prediction of the real motivation value in a specific fitness app is more difficult. This may be due to the concrete implementation of the elements in the second study, and due to the general problem of (theoretically) rating gamification elements without having experienced them in a real application.

Paper Nr: 38
Title:

A Knowledge-based Clinical Decision Support System for Headache Disorders Management

Authors:

Maria C. Groccia, Rosita Guido, Domenico Conforti and Rosario Iannacchero

Abstract: Headache is one of the most common neurological problems faced by General Practitioners (GPs) and neurologists. Most of GPs find the diagnosis of headache rather difficult: paper-based guidelines are long and the diagnostic criteria are complex. Thus, many headache patients do not have an early accurate diagnosis of headaches’ type and an appropriate treatment. In order to overcome this burden, we present a knowledge-based Clinical Decision Support System (CDSS) specifically devoted to support GPs in the headache diagnosis and in the appropriate selection of the diagnostic-therapeutic path. The proposed CDSS has been designed and developed based on internationally validated guidelines and clinical protocols. The knowledge base contains the medical-clinical knowledge appropriately formalized in several set of rules. Communication interfaces compliant with HL7 DSS (Health Level Seven Decision Support Service) international standard were developed enabling interoperation with other healthcare applications. The CDSS has been tested and assessed in the GPs’ daily practice of the Calabria Cephalalgic Network. During the evaluation period, a reduced number of requests for neurological visits and unnecessary and expensive instrumental examinations was registered. The results obtained from the evaluation period demonstrate that the CDSS turns out to be effective in the management of headache patients.

Paper Nr: 47
Title:

Computational Model for Changing Sedentary Behavior through Cognitive Beliefs and Introspective Body-feelings

Authors:

Fawad Taj, Nimat Ullah and Michel Klein

Abstract: Sedentary behavior has emerged as a serious risk factor for numerous health outcomes. However, little work has been done to approach the problem through social-cognitive theories. In this study, a network model has been proposed for sedentary behavior intervention based on Influential determinants from major social-cognitive theories i.e., theory of planned behavior and health-belief model. Accounting for these determinants means that we are influencing behavior with a peripheral route, for which we included the somatic markers as a body-feelings in the model. An effective behavior change techniques from literature are used to affect these determinants to change the sedentary behavior. The model has been mathematically represented and simulated using a network-oriented modelling technique for an office employee.

Paper Nr: 52
Title:

A Discrete SIR Model with Spatial Distribution on a Torus for COVID-19 Analysis using Local Neighborhood Properties

Authors:

Reinhard Schuster, Klaus-Peter Thiele, Thomas Ostermann and Martin Schuster

Abstract: The ongoing COVID-19 pandemic threatens the health of humans, causes great economic losses and may disturb the stability of the societies. Mathematical models can be used to understand aspects of the dynamics of epidemics and to increase the chances of control strategies. We propose a SIR graph network model, in which each node represents an individual and the edges represent contacts between individuals. For this purpose, we use the healthy S (susceptible) population without immune behavior, two I-compartments (infectious) and two R-compartments (recovered) as a SIR model. The time steps can be interpreted as days and the spatial spread (limited in distance for a singe step) shell take place on a 200 by 200 torus, which should simulate 40 thousand individuals. The disease propagation form S to the I-compartment should be possible on a k by k square (k=5 in order to be in small world network) with different time periods and strength of propagation probability in the two I compartments. After the infection, an immunity of different lengths is to be modeled in the two R compartments. The incoming constants should be chosen so that realistic scenarios can arise. With a random distribution and a low case number of diseases at the beginning of the simulation, almost periodic patterns similar to diffusion processes can arise over the years. Mean value operators and the Laplace operator on the torus and its eigenfunctions and eigenvalues are relevant reference points. The torus with five compartments is well suited for visualization. Realistic neighborhood relationships can be viewed with a inhomogeneous graph theoretic approach, but they are more difficult to visualize. Superspreaders naturally arise in inhomogeneous graphs: there are different numbers of edges adjacent to the nodes and should therefore be examined in an inhomogeneous graph theoretical model. The expected effect of corona control strategies can be evaluated by comparing the results with various constants used in simulations. The decisive benefit of the models results from the long-term observation of the consequences of the assumptions made, which can differ significantly from the primarily expected effects, as is already known from classic predator-prey models.

Paper Nr: 55
Title:

Disambiguating Clinical Abbreviations using Pre-trained Word Embeddings

Authors:

Areej Jaber and Paloma Martínez

Abstract: Abbreviations are broadly used in clinical texts and most of them have more than one meaning which makes them highly ambiguous. Determining the right sense of an abbreviation is considered a Word Sense Disambiguation (WSD) task in clinical natural language processing (NLP). Many approaches are applied to disambiguate abbreviations in clinical narrative. However, supervised machine learning approaches are studied in this field extensively and have proven a good performance at tackling this problem. We have investigated four strategies that integrate pre-trained word embedding as features to train two supervised machine learning models: Support Vector Machines (SVM) and Naive Bayes (NB). Our training features include information of the context of target abbreviation, which is applied on 500 sentences for each of the 13 abbreviations that have been extracted from public clinical notes data sets from the University of Minnesota-affiliated (UMN) Fairview Health Services in the Twin Cities. Our results showed that SVM performs better than NB in all four strategies; the highest accuracy being 97.08% using a pre-trained model trained from Wikipedia, PubMed and PMC (PubMedCentral) texts.

Paper Nr: 59
Title:

Epidemiological and Prognostic Factors Related to COVID-19 in Primary Care in a Municipality in Southern Brazil

Authors:

Luís R. Coutinho, Lecian C. Lopes, Henrique D. Neves, Hidelbrando Rodrigues, Sibele H. Costa and Fábio Sensever

Abstract: The COVID-19 pandemic crisis led to a reflection on health systems in general, addressing the importance not only of equitable, but also universal, care to populations in a global context. Numerous ways to identify and propose alternatives to COVID-19 issues are enabled by technological advances, specifically those applied to health. This study aimed to investigate COVID-19-related epidemiological and prognostic factors remotely in primary care in the Brazilian public health system. The sample consisted of 77 users of primary care who had records of respiratory symptoms during the pandemic, aged between 5 and 83 years. The study was carried out in 07 Health Centers in Porto Belo (SC), Brazil. Data analysis and subsequent evaluation found a possible demand with symptoms and sequelae of COVID-19 that was shown mainly by users who reported difficulty or “tiredness” in basic Activities of Daily Living (ADLs) and Instrumental Activities of Daily Living (IADLs), which may result in patients with chronic muscle fatigue and dyspnea as a future demand in the municipality.

Paper Nr: 68
Title:

Virtual Reality and Serious Games for Stress Reduction with Application in Work Environments

Authors:

I. Ladakis, V. Kilintzis, D. Xanthopoulou and I. Chouvarda

Abstract: This paper proposes a VR – based gamified approach aiming to reduce stress levels in work environments. This idea employs a standalone VR headset for immersion in a peaceful virtual environment, guided deep breathing exercises, sensing of heart rate and electrodermal activity for the estimation of stress, and feedback to the VR environment. For the qualitative and quantitative assessment of stress levels two sensors were used, Scosche Rhythm+ for monitoring heart rate signal (HR) and Moodmetric Ring for monitoring the EDA (electrodermal activity) signal. As a preliminary work, a series of stress-inducing batteries were created, including different stressful conditions that may resemble real life conditions (e.g., at work). Experiments with volunteers were conducted, to investigate the stress response before and during a stressful situation, as well as after VR-based relaxation. The signal fluctuation over time and the correlation between HR/EDA signals were explored towards selecting the optimal metric for the representation of the real stress level. The results of this preliminary study are relevant for the timely estimation of stress levels and the provision of a simple and useful tool for the immediate decrease of stress in various real-life environments such as work environments.

Paper Nr: 69
Title:

Using Data Analytics to Strengthen Monitoring and Surveillance of Routine Immunization Coverage for Children under One Year in Uganda

Authors:

Bartha A. Nantongo, Josephine Nabukenya and Peter Nabende

Abstract: Immunization coverage is a traditional key performance indicator that enables stakeholders to monitor child health, investigate gaps, and take remedial actions. It is continuously challenged by validity due to the neglect of unstructured data and process indicators that track small changes/milestones. While empirical evidence indicates digitalized immunization systems establish coverage from structured data, renowned administrative and household survey estimates are often inaccurate/untimely. Government instituted awareness, accessibility, and results-based performance approaches, but stakeholders are challenged by accurate monitoring of performance against Global Vaccination Action Plan coverage targets. This heightens inappropriate strategy implementation leading to persistent low coverage and declining trends. There is scanty literature substantiating the essence of comprehensive immunization indicators in monitoring evidence-based and timely interventions. For this reason, health workers failed to appreciate immunization process indicators and monitoring role. The study aims at developing a real-time immunization coverage monitoring framework that supports evidence-based strategy implementation using prescriptive analytics. The envisaged artifact analyzes a variety of data and monitors immunization performance against comprehensive indicators. It is a less resource-demanding strategy that prompts accurate and real-time insights to support intervention implementation decisions. This study will follow an explanatory research approach by first collecting quantitative data and later qualitative for in-depth analysis.

Paper Nr: 79
Title:

Usability and Security of Different Authentication Methods for an Electronic Health Records System

Authors:

Saptarshi Purkayastha, Shreya Goyal, Bolu Oluwalade, Tyler Phillips, Huanmei Wu and Xukai Zou

Abstract: We conducted a survey of 67 graduate students enrolled in the Privacy and Security in Healthcare course at Indiana University Purdue University Indianapolis. This was done to measure user preference and their understanding of usability and security of three different Electronic Health Records authentication methods: single authentication method (username and password), Single sign-on with Central Authentication Service (CAS) authentication method, and a bio-capsule facial authentication method. This research aims to explore the relationship between security and usability, and measure the effect of perceived security on usability in these three aforementioned authentication methods. We developed a formative-formative Partial Least Square Structural Equation Modeling (PLS-SEM) model to measure the relationship between the latent variables of Usability, and Security. The measurement model was developed using five observed variables (measures). - Efficiency and Effectiveness, Satisfaction, Preference, Concerns, and Confidence. The results obtained highlight the importance and impact of these measures on the latent variables and the relationship among the latent variables. From the PLS-SEM analysis, it was found that security has a positive impact on usability for Single sign-on and bio-capsule facial authentication methods. We conclude that the facial authentication method was the most secure and usable among the three authentication methods. Further, descriptive analysis was done to draw out the interesting findings from the survey regarding the observed variables.

Paper Nr: 84
Title:

WoundArch: A Hybrid Architecture System for Segmentation and Classification of Chronic Wounds

Authors:

Carlos D. F. da Rocha, Bruno M. Carvalho, Vítor G. Marques and Bruno S. Silva

Abstract: Every year, millions of people are affected by wounds worldwide. The wound treatment process is costly and requires the nurse to perform activities during patient care: tissue classification and calculation of the wound area. Thus, this work proposes to build a hybrid computer system with two configurations to support wound care. The first configuration uses a smartphone to perform the capture, segmentation and classification of the wound images. The other configuration has a client-server architecture, the images are captured and segmented in the application and sent, via the internet, to the web server, which is responsible for classifying the tissue of the wounds. The proposed methodology is the segmentation of images using the watershed algorithm and classification of tissues in Necrosis, Granulation or Crushing. fulfilled. To evaluate the application, experiments were performed with 20 images of wounds and the system was evaluated in two architectures: client and client-server. The results show that the client-server reached accelerations of up to 3.2 times in relation to the client-only architecture. The client-server architecture also saves energy and space in the client units, increasing the uptime of smartphones, in addition to reducing the storage load of the same.

Paper Nr: 86
Title:

On Open Workflows for Processing of Standardized Electroencephalography Data

Authors:

Roman Mouček and Filip Kupilík

Abstract: With increasing amounts of experimental data, openness, fairness, and reproducibility of scientific experimental work have become important factors for researchers, journals and funding bodies. However, these kinds of challenges are not easily and directly achievable. The goal of this paper is to contribute to these efforts by introducing advances in building more mature lifecycle of electroencephalography/event-related potential data. The progressive data standardization initiatives, data formats, and trends in using machine and deep learning methods for processing of domain data are described and discussed. An open processing workflow based on the analysis of current software tools for preprocessing, processing and classification of electroencephalography/event-related potential data is proposed, implemented and verified on a publicly available dataset.

Paper Nr: 89
Title:

Ultra-Wideband Radar Detection of Breathing Rate: A Comparative Evaluation

Authors:

Nicole Buckingham and Denis Gračanin

Abstract: The cost of a medical grade breathing rate monitors can be prohibitive. However, commodity ultra-wideband (UWB) radar based device can be used to detect breathing rate for health monitoring applications in homes. We identified several research challenges, including high cost and functional limitations based on the user’s location, orientation, and movement, as well as dependency on system placement and vulnerabilities in signal processing methods. We performed a comparative evaluation for a commodity UWB radar based device, Walabot, to determine its feasibility for health monitoring applications. The data was processed using two breathing rate derivation techniques: Fast Fourier Transformation (FFT) and Peak Detection. The results support feasibility of Walabot as a commodity breathing rate monitor for health monitoring in homes.

Paper Nr: 92
Title:

Digital Inclusion of Nursing Home Residents: A Usability Evaluation of the Digital Kiosk siosLIFE™

Authors:

Carla V. Leite, Daniel Carvalho, Ivone Almeida, Sofia Nunes and Ana M. Almeida

Abstract: The fast demographic ageing and technological progress are leading to a greater demand to develop digital solutions that can foster communication, information, and socialization of the elderly population. In the past decade, kiosks have been used to prevent digital exclusion and to promote the quality of life of this age group. This paper analyses siosLIFE™, a digital kiosk that is gradually getting attention from the public. The methodology adopted consisted in a usability test with guided tasks, using a cognitive walkthrough and think aloud protocol, with participants being residents from a nursing home. The results show that siosLIFE™ complies with some usability recommendations, but there are several improvements regarding the interface, contents, integration of support systems, and assistive technologies that can be made.

Paper Nr: 98
Title:

A Survey of Survival Analysis Techniques

Authors:

George Marinos and Dimosthenis Kyriazis

Abstract: Survival analysis is a branch of statistics for analyzing the expected duration of the time until the event of interest happens. It is not only applicable to biomedical problems but it can be widely used in almost every domain since there is a relevant data structure available. Recent studies have shown that it is a powerful approach for risk stratification. Since it is a well established statistical technique, there have been several studies that combine survival analysis with machine learning algorithms in order to obtain better performances. Additionally in the machine learning scientific field the usage of different data modalities has been proven to enhance the performance of predictive models. The majority of the scientific outcomes in the survival analysis domain have focused on modeling survival data and building robust predictive models for time to event estimation. Clustering based on risk-profiles is partly under-explored in machine learning, but is critical in applications domains such as clinical decision making. Clustering in terms of survivability is very useful when there is a need to identify unknown sub-populations in the overall data. Such techniques aim for identification of clusters whose lifetime distributions significantly differs, which is something that is not able to be done by applying traditional clustering techniques. In this survey we present research studies in the aforementioned domain with an emphasis on techniques for clustering censored data and identifying various risk level groups.

Paper Nr: 104
Title:

Epinoter: A Natural Language Processing Tool for Epidemiological Studies

Authors:

Weisong Liu, Fei Li, Yonghao Jin, Edgard Granillo, Jorge Yarzebski, Wenjun Li and Hong Yu

Abstract: Clinical epidemiological studies often rely on manual processing of electronic health records (EHRs). Manual chart review is, however, costly, time consuming, and difficult to scale up. Natural language processing (NLP) is a promising alternative to this process since it can automatically extract and process large amounts of EHR data. To date, only a few reported integrations of NLP systems into epidemiological studies. In this study, we report the development, integration, and evaluation of Epinoter, a semi-automated NLP annotation system for use in epidemiologic studies. This new system has been integrated into the population-based Worcester Heart Attack Study, to streamline the review of hospital medical records by automatically detecting variables of interest associated with hospitalization for acute myocardial infarction (AMI). Epinoter improves the efficiency of reviewing patients’ electronic charts by automatically identifying pertinent AMI elements from structured EHR data. In addition, for unstructured EHR notes, Epinoter automatically highlights the most relevant words and sentences for key AMI elements, removes unimportant information such as redundant paragraphs and common templated sentences that are unrelated to AMI, and validates important AMI data elements. Epinoter achieved 63% sensitivity and 90% specificity. Our results demonstrate the effectiveness of Epinoter for epidemiological studies of AMI.