HEALTHINF 2020 Abstracts


Full Papers
Paper Nr: 13
Title:

A Hybrid Visual Analytics Framework to Evaluate Trauma Incidences and Enhance Patient Care

Authors:

Waqar Haque, Jordan Oliver, Sonal Bajaj and Navjot Kaur

Abstract: Large volumes of data has been collected on unintentional injuries and mortality. Besides years of life lost, traumatic injuries account for a significant portion of healthcare expenditure. With intelligent visual analytics, the collected data can be used for informed decision making and resource allocation. A multi-dimensional online analytical processing (OLAP) cube has been developed using data from BC Trauma Registry (BCTR) and Discharge Abstract Database (DAD). We propose a comprehensive framework which uses the OLAP cube, a web-based data entry platform populating a standalone database, filters consistent with Accreditation Canada (AC) measures, AC inclusion/exclusion criteria, and tools which render reports from historical and operational perspectives. These reports are grouped in higher level categories with intuitive drill-down capabilities for navigating at finer granularity along multiple dimensions. Access control for data entry is enabled with provisions for nested groups.

Paper Nr: 15
Title:

A Feature-based Approach for Identifying Soccer Moves using an Accelerometer Sensor

Authors:

Omar Alobaid and Lakshmish Ramaswamy

Abstract: During the past decade, Human Activity Recognition (HAR) systems have been an evolving topic due to the popularity of smart devices. Recognizing soccer moves in real-time is an important research problem that has not yet been studied thoroughly in the literature. In contrast to daily physical activities, recognizing soccer moves poses a set of unique challenges, such as pattern irregularity and body positions when performing these moves. In this paper, our goal is to recognize soccer moves in real-time by utilizing accelerometer data. We explore three different feature-based algorithms: Time Series Forest, Fast Shapelets, and Bag-of-SFA-Symbols. We also examine different factors that can affect the performance of these algorithms, such as parameter tuning and accelerometer axis elimination. Additionally, we introduce a novel collaborative model consisting of the above-mentioned algorithms in a majority voting mechanism to further enhance the performance of the system. We also add a light-weight classifier to act as a tie breaker in case of disagreement between the classifiers. We experimentally choose the right parameters to reduce the training time drastically without forfeiting the level of accuracy. Our collaborative model outperforms the single model by 2% to reach 84% in accuracy with a decrease in the training time by one order of magnitude.

Paper Nr: 17
Title:

Deep Learning from Heterogeneous Sequences of Sparse Medical Data for Early Prediction of Sepsis

Authors:

Mahbub U. Alam, Aron Henriksson, John K. Valik, Logan Ward, Pontus Naucler and Hercules Dalianis

Abstract: Sepsis is a life-threatening complication to infections, and early treatment is key for survival. Symptoms of sepsis are difficult to recognize, but prediction models using data from electronic health records (EHRs) can facilitate early detection and intervention. Recently, deep learning architectures have been proposed for the early prediction of sepsis. However, most efforts rely on high-resolution data from intensive care units (ICUs). Prediction of sepsis in the non-ICU setting, where hospitalization periods vary greatly in length and data is more sparse, is not as well studied. It is also not clear how to learn effectively from longitudinal EHR data, which can be represented as a sequence of time windows. In this article, we evaluate the use of an LSTM network for early prediction of sepsis according to Sepsis-3 criteria in a general hospital population. An empirical investigation using six different time window sizes is conducted. The best model uses a two-hour window and assumes data is missing not at random, clearly outperforming scoring systems commonly used in healthcare today. It is concluded that the size of the time window has a considerable impact on predictive performance when learning from heterogeneous sequences of sparse medical data for early prediction of sepsis.

Paper Nr: 23
Title:

The Development and Psychometric Assessment of Medication Literacy Scale for Hypertensive Patients

Authors:

Zhuqing Zhong, Aijing Luo, Wenzhao Xie, Siqing Ding, Shuangjiao Shi, Yinglong Duan and Feng Zheng

Abstract: Objective: To develop the medication literacy scale for patients with hypertension, and to test the reliability and validity of the scale. Methods The initial draft of the scale was formulated based on a theoretical framework of medication literacy with four domains of knowledge, attitude, skill and practice, and developed through procedures of literature review, expert meetings and consultations, patient interviews and focus group discussions. In this study, 260 patients with hypertension in Changsha city of China were selected to conduct a pilot survey. After item selection by a series of statistical analysis method and item re-wording according to patients’ feedback, the scale was revised to form a formal investigation scale with four domains and 37 items. A formal investigation was carried out on 650 patients with hypertension selected purposively in a tertiary general hospital and two community health service centers in Changsha city. The reliability and validity of the scale were analyzed. Results: Finally, the formal scale consists of four dimensions on knowledge, attitude, practice and skills, 11 loading factors and 37 items in total. S-CVI of the scale was 0.968, and the I-CVI for each item ranged from 0.833 to 1.000, indicating good and acceptable content and face validity. The Cronbach’s α coefficient was 0.849 for the overall scale and ranged from 0.744 to 0.783 for 4 dimensions. The Pearson correlation coefficient between each of the four dimension and the total scale was 0.530-0.799. Besides, the Pearson correlation coefficient among each dimension of the scale ranged from 0.157 to 0.439. The split-half reliability coefficient was 0.893 for the total scale and ranged from 0.793 to 0.872 for four dimensions. The test-retest reliability coefficient of the total scale was 0.968 and ranged from 0.880 to 0.959 for four dimensions. 11 common loading factors were extracted through exploratory factor analysis, and the cumulative variance contribution rate of individual domains were 56.111%-64.419%. The confirmatory factor analysis showed the fit indices of the four-dimension 11-factor model as follows (2/df=2.629,GFI=0.804,AGFI=0.777,RMR=0.012,IFI=0.746,RMSEA=0.066,PNFI=0.599,PCFI=0.689), which indicated good model fit. Conclusions: The medication literacy scale for hypertensive patients has good reliability and validity, which is suitable and acceptable for evaluating the medication literacy level of hypertension patients in China. In the future, English translation of this scale is required, so that this scale can be further validated and applied worldwide.

Paper Nr: 25
Title:

Resolving Differences of Opinion between Medical Experts: A Case Study with the IS-DELPHI System

Authors:

Derek Sleeman, Kiril Kostadinov, Laura Moss and Malcolm Sim

Abstract: Knowledge intensive clinical systems, as well as machine learning algorithms, have become more widely used over the last decade or so. These systems often need access to sizable labelled datasets which could be more useful if their instances are accurately labelled / annotated. A variety of approaches, including statistical ones, have been used to label instances. In this paper, we discuss the use of domain experts, in this case clinicians, to perform this task. Here we recognize that even highly rated domain experts can have differences of opinion on certain instances; we discuss a system inspired by the Delphi approaches which helps experts resolve their differences of opinion on classification tasks. The focus of this paper is the IS-DELPHI tool which we have implemented to address the labelling issue; we report its use in a medical domain in a study involving 12 Intensive Care Unit clinicians. The several pairs of experts initially disagreed on the classification of 11 instances but as a result of using IS-DELPHI all those disagreements were resolved. From participant feedback (questionnaires), we have concluded that the medical experts understood the task and were comfortable with the functionality provided by IS-DELPHI. We plan to further enhance the system’s capabilities and usability, and then use IS-DELPHI, which is a domain independent tool, in a number of further medical domains.

Paper Nr: 28
Title:

Extracting Behavioral Determinants of Health from Electronic Health Records: Classifying Yoga Mentions in the Clinic

Authors:

Nadia M. Penrod, Selah Lynch and Jason H. Moore

Abstract: Behavior-based interventions can prevent and/or treat many common chronic diseases, but few clinical research studies incorporate behavioral data. Collecting behavioral data on a large-scale is time-consuming and expensive. Fortunately, electronic health records (EHRs) are an incidental source of population-level behavioral data captured in clinical narratives as unstructured, free text. Here, we developed and evaluated three supervised text classification models for stratifying clinical chart notes based on use of yoga, a behavioral determinant of health that is linked to stress-management and the prevention and treatment of chronic disease. We demonstrate that yoga can be extracted from the EHR and classified into meaningful use cases for inclusion in clinical research.

Paper Nr: 32
Title:

Data Mining in Clinical Trial Text: Transformers for Classification and Question Answering Tasks

Authors:

Lena Schmidt, Julie Weeds and Julian T. Higgins

Abstract: This research on data extraction methods applies recent advances in natural language processing to evidence synthesis based on medical texts. Texts of interest include abstracts of clinical trials in English and in multilingual contexts. The main focus is on information characterized via the Population, Intervention, Comparator, and Outcome (PICO) framework, but data extraction is not limited to these fields. Recent neural network architectures based on transformers show capacities for transfer learning and increased performance on downstream natural language processing tasks such as universal reading comprehension, brought forward by this architecture’s use of contextualized word embeddings and self-attention mechanisms. This paper contributes to solving problems related to ambiguity in PICO sentence prediction tasks, as well as highlighting how annotations for training named entity recognition systems are used to train a high-performing, but nevertheless flexible architecture for question answering in systematic review automation. Additionally, it demonstrates how the problem of insufficient amounts of training annotations for PICO entity extraction is tackled by augmentation. All models in this paper were created with the aim to support systematic review (semi)automation. They achieve high F1 scores, and demonstrate the feasibility of applying transformer-based classification methods to support data mining in the biomedical literature.

Paper Nr: 37
Title:

Adaptative Clinical Decision Support System using Machine Learning and Authoring Tools

Authors:

Jon Kerexeta, Jordi Torres, Naiara Muro, Kristin Rebescher and Nekane Larburu

Abstract: Clinical Decision Support Systems (CDSS) offer the potential to improve quality of clinical care and patients’ outcomes while reducing medical errors and economic costs. The development of these systems results difficult since (i) generating the knowledge base that CDSS use to evaluate clinical data requires technical and clinical knowledge, and (ii) usually the reasoning process of CDSS is difficult to understand for clinicians leading to a low adherence to the recommendations provided by these systems. Hereafter, to address these issues, we propose a web-based platform, named Knowledge Generation Tool (KGT), which (i) enables clinicians to take an active role in the creation of the CDSSs in a simple way, and (ii) clinicians’ involvement can turn in an improvement of the model predictor capabilities, while their comprehension of the reasoning process of the CDSS is increased. The KGT consist on three main modules: DT building, which implements machine learning methods to extract automatically decision trees (DTs) from clinical data frames; an authoring tool (AT), which enables the clinicians to modify the DT with their expert knowledge, and the DT testing, which allows to test any DT, being able to test objectively any modification made by clinician’s expert knowledge.

Paper Nr: 42
Title:

SugarArray: A User-centred-designed Platform for the Analysis of Lectin and Glycan Microarrays

Authors:

Aurora Sucre, Raquel Pazos, Niels-Christian Reichardt and Alba Garín-Muga

Abstract: Glycan and lectin microarrays are two arising technologies, very important to the glycomics field. Glycomics is the science that focuses on defining the structures and functions of carbohydrates in nature. These microarrays provide information regarding the interactions between specific carbohydrates and proteins, and it has many applications in clinical and research settings. Nevertheless, the availability of analytical software for these types of arrays is very limited, so researchers usually perform data processing and analytical pipelines manually, which is very time consuming and prone to error. SugarArray was born as a user-friendly and intuitive stand-alone solution that process the intensity data generated from glycan or lectin array studies, and displays the results to the user in an understandable manner. The solution also allows the users to manage the data as needed, create data plots and automatically generate reports. This tool was intended to simplify the processing steps of the analytical pipeline, so the users can focus on what really matters: understanding the results.

Paper Nr: 45
Title:

Applying Machine Learning on Patient-Reported Data to Model the Selection of Appropriate Treatments for Low Back Pain: A Pilot Study

Authors:

Wendy –. d’Hollosy, Lex van Velsen, Mannes Poel, Catharina M. Groothuis-Oudshoorn, Remko Soer, Patrick Stegeman and Hermie Hermens

Abstract: The objective of this pilot study was to determine whether machine learning can be applied on patient-reported data to model decision-making on treatments for low back pain (LBP). We used a database of a university spine centre containing patient-reported data from 1546 patients with LBP. From this dataset, a training dataset with 354 features (input data) was labelled on treatments (output data) received by these patients. For this pilot study, we focused on two treatments: pain rehabilitation and surgery. Classification algorithms in WEKA were trained, and the resulting models were validated during 10-fold cross validation. Next to this, a test dataset was constructed - containing 50 cases judged on treatments by 4 master physician assistants (MPAs) - to test the models with data not used for training. We used prediction accuracy and average area under curve (AUC) as performance measures. The interrater agreement among the 4 MPAs was substantial (Fleiss Kappa 0.67). The AUC values indicated small to medium (machine) learning effects, meaning that machine learning on patient-reported data to model decision-making processes on treatments for LBP seems possible. However, model performances must be improved before these models can be used in real practice.

Paper Nr: 48
Title:

Health Information Systems for Clients with Mild Intellectual and Developmental Disability: A Framework

Authors:

Muneef Alshammari, Owen Doody and Ita Richardson

Abstract: Persons with intellectual and developmental disability (IDD) remain among the most vulnerable members of society and frequently face numerous barriers accessing healthcare services. Following our recent literature review identifying needs for persons with IDD, we propose that frameworks can be useful to identify the key components of person-centred health information. These will ultimately support the building of relevant Health Information Systems. This paper presents the initial development and content of the Person-Centred Health Information framework (PCHI) developed to support persons with mild IDD. PCHI is based on the Information-Motivation-Behavioral skills (IMB) model and its use in the design and development of Health Information Systems has the potential to improve health access and outcomes for persons with mild IDD.

Paper Nr: 50
Title:

Explain Yourself: A Semantic Annotation Framework to Facilitate Tagging of Semantic Information in Health Smart Homes

Authors:

Bastian Wollschlaeger, Elke Eichenberg and Klaus Kabitzsch

Abstract: Health Smart Homes (HSH) are a key concept in future personalized health care. However, with an abundance of heterogeneous assistance components available, the design process of HSHs is becoming increasingly complex and needs to rely on computer-based support. As a key prerequisite, formal models of information semantics are required for component selection and early-on interoperability assessment. This paper proposes a flexible and extensible framework of semantic annotations that copes with the interdisciplinary nature of HSH and can be used to incorporate a formal model of semantics at component interfaces. The framework consists of a taxonomy of semantic annotations (tags) and their relationships and is developed using a well-established taxonomy creation method. The tags address several independent aspects of semantics, increasing the expressiveness of information semantics. With this valuable new level of detail for information, component models can be semantically enriched, in turn enabling computer-based design algorithms to carry out the complex design tasks in the future.

Paper Nr: 58
Title:

Evaluating Cross-lingual Semantic Annotation for Medical Forms

Authors:

Ying-Chi Lin, Victor Christen, Anika Groß, Toralf Kirsten, Silvio D. Cardoso, Cédric Pruski, Marcos Da Silveira and Erhard Rahm

Abstract: Annotating documents or datasets using concepts of biomedical ontologies has become increasingly important. Such ontology-based semantic annotations can improve the interoperability and the quality of data integration in health care practice and biomedical research. However, due to the restrictive coverage of non-English ontologies and the lack of comparably good annotators as for English language, annotating non-English documents is even more challenging. In this paper we aim to annotate medical forms in German language. We present a parallel corpus where all medical forms are in both German and English languages. We use three annotators to automatically generate annotations and these annotations are manually verified to construct an English Silver Standard Corpus (SSC). Based on the parallel corpus of German and English documents and the SSC, we evaluate the quality of different annotation approaches, mainly 1) direct annotation using German corpus and German ontologies and 2) integrating machine translators to translate German corpus and annotate the translated corpus with English ontologies. The results show that using German ontologies only produces very restricted results, whereas translation achieves better annotation quality and is able to retain almost 70% of the annotations.

Paper Nr: 59
Title:

Combining Rhythmic and Morphological ECG Features for Automatic Detection of Atrial Fibrillation

Authors:

Gennaro Laudato, Franco Boldi, Angela R. Colavita, Giovanni Rosa, Simone Scalabrino, Paolo Torchitti, Aldo Lazich and Rocco Oliveto

Abstract: Atrial Fibrillation (AF) is a common cardiac disease which can be diagnosed by analyzing a full electrocardiogram (ECG) layout. The main features that cardiologists observe in the process of AF diagnosis are (i) the morphology of heart beats and (ii) a simultaneous arrhythmia. In the last decades, a lot of effort has been devoted for the definition of approaches aiming to automatic detect such a pathology. The majority of AF detection approaches focus on R-R Intervals (RRI) analysis, neglecting the other side of the coin, i.e., the morphology of heart beats. In this paper, we aim at bridging this gap. First, we present some novel features that can be extracted from an ECG. Then, we combine such features with other classical rhythmic and morphological features in a machine learning based approach to improve the detection accuracy of AF events. The proposed approach, namely MORPHYTHM, has been validated on the Physionet MIT-BIH AF Database. The results of our experiment show that MORPHYTHM improves the classification accuracy of AF events by correctly classifying about 4,400 additional instances compared to the best state of the art approach.

Paper Nr: 60
Title:

Privacy-preserving Metrics for an mHealth App in the Context of Neuropsychological Studies

Authors:

Alexander Gabel, Funda Ertas, Michael Pleger, Ina Schiering and Sandra V. Müller

Abstract: The potential of smart devices as smartphones, smart watches and wearables in healthcare and rehabilitation, so-called mHealth applications, is considerable. It is especially interesting, that these devices accompany patients during their normal life. Hence they are able to track activities and support users in activities of daily life. But beside the benefits for patients, mHealth applications also constitute a considerable privacy and security risk. The central question investigated here is how data about the usage of mobile applications in empirical studies with mHealth technologies can be collected in a privacy-friendly way based on the ideas of Privacy by Design. The context for the proposed approach are neuropsychological studies where a mobile application for Goal Management Training, a therapy for executive dysfunctions, is investigated. There a privacy-friendly concept for collecting data about the usage of the app based on metrics which are derived from research questions is proposed. The main ideas underlying the proposed concept are a decentralized architecture, where only aggregated data is gathered for the study, and a consequent data minimization approach.

Paper Nr: 62
Title:

Receptivity of an AI Cognitive Assistant by the Radiology Community: A Report on Data Collected at RSNA

Authors:

Karina Kanjaria, Anup Pillai, Chaitanya Shivade, Marina Bendersky, Vandana Mukherjee and Tanveer Syeda-Mahmood

Abstract: Due to advances in machine learning and artificial intelligence (AI), a new role is emerging for machines as intelligent assistants to radiologists in their clinical workflows. But what systematic clinical thought processes are these machines using? Are they similar enough to those of radiologists to be trusted as assistants? A live demonstration of such a technology was conducted at the 2016 Scientific Assembly and Annual Meeting of the Radiological Society of North America (RSNA). The demonstration was presented in the form of a question-answering system that took a radiology multiple choice question and a medical image as inputs. The AI system then demonstrated a cognitive workflow, involving text analysis, image analysis, and reasoning, to process the question and generate the most probable answer. A post demonstration survey was made available to the participants who experienced the demo and tested the question answering system. Of the reported 54,037 meeting registrants, 2,927 visited the demonstration booth, 1,991 experienced the demo, and 1,025 completed a post-demonstration survey. In this paper, the methodology of the survey is shown and a summary of its results are presented. The results of the survey show a very high level of receptiveness to cognitive computing technology and artificial intelligence among radiologists.

Paper Nr: 66
Title:

Improving Mental Health using Machine Learning to Assist Humans in the Moderation of Forum Posts

Authors:

Dong Wang, Julie Weeds and Ian Comley

Abstract: This work investigates the potential for the application of machine learning and natural language processing technology in an online application designed to help teenagers talk about their mental health issues. Specifically, we investigate whether automatic classification methods can be applied with sufficient accuracy to assist humans in the moderation of posts and replies to an online forum. Using real data from an existing application, we outline the specific problems of lack of data, class imbalance and multiple rejection reasons. We investigate a number of machine learning architectures including a state-of-the-art transfer learning architecture, BERT, which has performed well elsewhere despite limited training data, due to its use of pre-training on a very large general corpus. Evaluating on real data, we demonstrate that further large performance gains can be made through the use of automatic data augmentation techniques (synonym replacement, synonym insertion, random swap and random deletion). Using a combination of data augmentation and transfer learning, performance of the automatic classification rivals human performance at the task, thus demonstrating the feasibility of deploying these techniques in a live system.

Paper Nr: 67
Title:

MIPHAS: Military Performances and Health Analysis System

Authors:

Gennaro Laudato, Giovanni Rosa, Simone Scalabrino, Jonathan Simeone, Francesco Picariello, Ioan Tudosa, Luca De Vito, Franco Boldi, Paolo Torchitti, Riccardo Ceccarelli, Fabrizio Picariello, Luca Torricelli, Aldo Lazich and Rocco Oliveto

Abstract: In the last few years wearable devices are becoming always more important. Their usefulness mainly lies in the continuous monitoring of vital parameters and signals, such as electrocardiogram. However, such a monitoring results in an enormous amount of data which cannot be precisely analyzed manually. This recalls the need of approaches and tools for the automatic analysis of acquired data. In this paper we present MIPHAS, a software system devised to meet this need in a well-defined context: the monitoring of athletes during sport activities. MIPHAS is a system composed of several components: a smart t-shirt, an electronic component, a web application, a mobile APP and an advanced decision support system based on machine learning techniques. This latter is the core component of MIPHAS dedicated to the automatic detection of potential anomalies during the monitoring of vital parameters.

Paper Nr: 68
Title:

Food Data Integration by using Heuristics based on Lexical and Semantic Similarities

Authors:

Gorjan Popovski, Gordana Ispirova, Nina Hadzi-Kotarova, Eva Valenčič, Tome Eftimov and Barbara K. Seljak

Abstract: With the rapidly growing food supply in the last decade, vast amounts of food-related data have been collected. To make this data inter-operable and equipped for analyses involving studying relations between food, as one of the main environmental and health outcomes, data coming from various data sources needs to be normalized. Food data can have varying sources and formats (food composition, food consumption, recipe data), yet the most familiar type is food product data, often misinterpreted due to marketing strategies of different producers and retailers. Several recent studies have addressed the problem of heterogeneous data by matching food products using lexical similarity between their English names. In this study, we address this problem, while considering a non-English, low researched language in terms of natural language processing, i.e. Slovenian. To match food products, we use our previously developed heuristic based on lexical similarity and propose two new semantic similarity heuristics based on word embeddings. The proposed heuristics are evaluated using a dataset with 438 ground truth pairs of food products, obtained by matching their EAN barcodes. Preliminary results show that the lexical similarity heuristic provides more promising results (75% accuracy), while the best semantic similarity model yields an accuracy of 62%.

Paper Nr: 86
Title:

Fast and Realistic Approach to Virtual Muscle Deformation

Authors:

Martin Cervenka and Josef Kohout

Abstract: This paper describes a real-time simulation of muscle movement that is based on inverse kinematics where bone placement is known apriori and muscle shape is calculated by a modified position-based dynamics (PBD) method. The method is comparable and competitive with the others, moreover, it is enhanced with some novel features like approach for respecting muscle anisotropy, really fast simplistic collision detection, etc. This real-time simulation presents visual plausibility of resulting muscle deformation in most cases.

Paper Nr: 93
Title:

Air Quality and Cause-specific Mortality in the United States: Association Analysis by Regression and CCA for 1980-2014

Authors:

Erin Teeple, Caitlin Kuhlman, Brandon Werner, Randy Paffenroth and Elke Rundensteiner

Abstract: Quantifying health effects resulting from environmental exposures is a complex task. Underestimation of exposure-outcome associations may occur due to factors such as data quality, jointly distributed spectra of possible effects, and uncertainty about exposure levels. Parametric methods are commonly used in population health research because parameter estimates, rather than predictive accuracy, are useful for informing regulatory policies. This project considers complementary approaches for capturing population-level exposure-outcome associations: multiple linear regression and canonical correlation analysis (CCA). We apply these methods for the task of characterizing relationships between air quality and cause-specific mortality. We first create a national air pollution exposures-mortality outcomes data set by integrating United States Environmental Protection Agency (EPA) annual summary county-level air quality measurements for the period 1980-2014 with age-adjusted gender- and cause-specific county mortality rates from the same time period published by the Institute for Health Metrics and Evaluation (IHME). Code for data integration is made publicly available. We examine our model parameter estimates together with air quality-mortality rate associations, revealing statistically significant correlations between air quality variations and variations in cause-specific mortality which are particularly apparent when CCA is applied to our population health data set.

Paper Nr: 95
Title:

Secure Audit in Support of an Adrenal Cancer Registry

Authors:

Anthony Stell, Vedant Chauhan and Richard Sinnott

Abstract: This paper describes the use of blockchain technology to ensure the integrity of data logs for a clinical registry, providing a technological means for a secure audit of that registry. The characteristics of a secure audit – tamper-resistance, verifiability, searchability and privacy – are described in their application to this registry, then an evaluation is performed detailing the use of blockchain to achieve these audit goals. The clinical registry tested – supporting ENSAT (the European Network for the Study of Adrenal Tumors) – is a production repository of clinical, phenotypic and genetic information about patients with adrenal cancer, a rare but often serious condition that affects approximately 1 in 600,000 of the world population. The registry is implemented using a standard n-tier web application, with a MySQL database back-end, and Java/JSP business logic and user interface. The information contributing to the full audit of data and usage of the registry, is captured in the application log-files, which are stored in two “mirrored” formats: ASCII text files compiled through the Java log4j project and in a MongoDB NoSQL database. Following a discussion of the relevant supporting features, the fully implemented solution – a private blockchain known as “ensatChain” – is evaluated for overall security, using the Microsoft “STRIDE” threat model.

Paper Nr: 107
Title:

Using a Self-Assessment Tool (SAT) to Review National Health Information Systems in Ireland

Authors:

Sarah Craig

Abstract: This paper presents a review of four national health information systems in Ireland using a nationally agreed self-assessment tool (SAT) developed in Ireland by the health information regulatory body there. The review was undertaken using documentary analysis of written materials about the systems from both primary and secondary data sources. The findings show high levels of compliance with the standards identified but that there is still some work to do to ensure that all aspects of the standards are met.

Paper Nr: 109
Title:

Enhancing Decision-making Systems with Relevant Patient Information by Leveraging Clinical Notes

Authors:

João R. Almeida, João F. Silva, Alejandro P. Sierra, Sergio Matos and José L. Oliveira

Abstract: Hospitalised patients suffering from secondary illnesses that require daily medication typically need personalised treatment. Although clinical guidelines were designed considering those circumstances, existing decision-support features fail in assimilating detailed relevant patient information, which opens up opportunities for systems capable of performing a real-time evaluation of such data against existing knowledge and providing recommendations during clinical treatments. In this paper, we present a proposal for a new feature to integrate with electronic health record (EHR) systems that enriches the health treatment process by automatically extracting information from patient medical notes and aggregating it in clinical protocols. Our goal is to leverage the historical component of the patient trajectory to improve clinical decision support systems performance.

Paper Nr: 116
Title:

A User Centred Approach in the Implementation of Mobile Marketing in Health Applications

Authors:

Tiago B. Fernandes and André Vasconcelos

Abstract: Medical appointments booking applications allow an easy way of accessing information about a healthcare provider like the possibility to see the CV of a medical health professional, an easy acquisition of a healthcare service or being able to see other patients reviews of the service. This paper describes the development of a mobile application, having mobile marketing as the main feature of it. This mobile application is an extension of a web based medical care appointment booking service. This research explores the various possibilities of mobile marketing by applying and reviewing User Interface (UI) and User Experience (UX) guidelines in order to provide the best user experience. The mobile marketing side of the application is used to offer patients suggestions for booking an appointment based on the location, medical history and recent searches on the application. Regarding the User Experience, several guidelines are assessed including the simplification of the medical appointment booking process, the presentation of an onboard screen for easier user interaction and all services before asking for a login. Through various test scenarios, using a focus group approach, the application presents satisfying user experience results, above the market average comparable applications.

Paper Nr: 120
Title:

Mining Patient Flow Patterns in a Surgical Ward

Authors:

Christoffer O. Back, Areti Manataki and Ewen Harrison

Abstract: Surgery is a highly critical and costly procedure, and there is an imperative need to improve the efficiency in surgical wards. Analyzing surgical patient flow and predicting cycle times of different peri-operative phases can help improve the scheduling and management of surgeries. In this paper, we propose a novel approach to mining temporal patterns of surgical patient flow with the use of Bayesian belief networks. We present and compare three classes of probabilistic models and we evaluate them with respect to predicting cycle times of individual phases of patient flow. The results of this study support previous work that surgical times are log-normally distributed. We also show that the inclusion of a clustering pre-processing step improves the performance of our models considerably.

Paper Nr: 122
Title:

Detailed Classification of Meal-related Activities from Eating Sound Collected in Free Living Conditions

Authors:

Archit Jain, Takumi Kondo, Haruka Kamachi, Anna Yokokubo and Guillaume Lopez

Abstract: Increasing the number of chews of each bite episode of a meal can help reduce obesity. Nevertheless, it is difficult for a person to keep track of his mastication rate without the help of an automatic mastication counting device. Such devices do exist, but they are big and non-portable and are not suitable for daily use. In our previous work, we proposed an optimization model for the classification of three meal-related activities, chewing, swallowing, and speaking activities from sound signals collected in free-living conditions with a cheap bone conduction microphone. To extract the number of chews per bite, it is necessary to differentiate the swallowing of food from the swallowing of drink. In this paper, we propose a new model that can not only classify speaking, chewing, and swallowing, but also differentiate whether swallowing is for food or drink, with an average accuracy of 96%.

Short Papers
Paper Nr: 3
Title:

Advanced Analytics to Predict Survivability of Breast Cancer Patients

Authors:

Sonal Bajaj and Waqar Haque

Abstract: A frequently asked question by cancer patients post-diagnosis is the lifespan they are left with. The oncologist’s response is generally based on past records of cancer patients with similar prognosis or by consulting other physicians and researchers working on comparable cases. Although careful prognosis is vital, it is difficult to predict accurate survival time of patients as survivability is based on many factors. Also, these predictions may not be accurate as the past records are not completely reliable and the prognosis from different oncologists are generally inconsistent. Further, existing repositories of data are not easily accessible and the stored formats are difficult to analyze. We propose an end-to-end process to build a model which predicts survival months of breast cancer patients. The predictive model is trained, tested and validated with different subsets of data. The modeling techniques used in this research are Neural Networks, CHAID, C&RT and an Ensemble of these techniques. The predictive model can also be used as a calculator which predicts survival months of a specific case.

Paper Nr: 4
Title:

A Novel Blockchain based Platform to Support Chronic Care Model Information Management

Authors:

Luigi Lella and Sergio Piersantelli

Abstract: Blockchain technology has been successfully used in many healthcare contexts, guaranteeing not only high security and privacy levels in clinical data management, but also the continuous updating of patient clinical pictures, to ensure the continuity of care and the reliability of data sources in statistical processing. These results are related to the peculiar features of this technology such as the distributed ledger, the chaincode, the encryption algorithms used to cypher information, the technological solutions used for block validation and the use of smart contracts. This article aims to present a possible solution based on blockchain technology to the problem of information management in the Chronic Care Model. The use of the blockchain makes it possible to create a patient-centred system that not only allows patients, or authorized people, to exercise a constant control over their health data, but it is also able to "contractualize" the agreements made in this regard together with the collection of consent for the processing of health data. The blockchain also allows the preparation of validated data sources for the subsequent statistical processing to update process and outcome indicators and the risk prospects related to the care pathways activated for patients suffering from chronic pathologies.

Paper Nr: 9
Title:

Analysis of the Use of Colour for Early Detection of Dementia

Authors:

Thomas Ostermann, Sibylle Robens, Petra Heymann, Sebastian Unger, Stephan Müller, Christoph Laske and Ulrich Elbing

Abstract: Cognitive visuo-constructive impairments, which can be detected by drawing tasks are early signs of Alzheimer’s disease (AD). Additionally, several studies revealed deficits in colour perception for patients with AD. In a former analysis of the impact of digital tree-drawing parameters on the screening of early dementia, a logistic regression revealed the number of colours together with the drawing velocity and the number of line widths changes as discrimination characteristics (ROC AUC=0.90, sensitivity=.86, specificity=0.82). To analyse the diagnostic importance of colour variations in drawings, a reanalysis of these data was done with 67 healthy subjects (25 females, mean age 66 ± 10 yrs.) and 56 subjects with early AD (40 females, mean age 73 ± 9 yrs.). The exclusion of colour variables resulted in a good discrimination of healthy and AD (ROC AUC=0.89, specificity=0.89) but in a reduction of sensitivity to .77 compared to the former model. This suggest that the analysis of colour variations in drawings has an important diagnostic impact.

Paper Nr: 11
Title:

A Responsive Web Application for the Improvement of Healthy Habits in the Child Population

Authors:

José A. Benítez-Andrades, Marta Martínez-Martínez, Isaías García-Rodríguez, Carmen Benavides, Carlos Fernández-San-Juan and Pilar Marqués-Sánchez

Abstract: The World Health Organization declared childhood obesity as a global epidemic, stressing the need for urgent action in order to face this serious public-health problem. This paper presents a pilot program based on a technological solution to improve the nutrition and physical activity habits in children. The solution is a multiplatform application that can be accessed both from computers and mobile devices. A study of current, similar solutions was carried out and a web application was designed and built, combining information on healthy nutrition and physical activity habits with the functionalities of an online social network. This application was used to carry out a pilot intervention with children from the first and second years of Compulsory Secondary Education (CSE) at a school in Spain. After the pilot program, feedback was received from the users, obtaining valuable information for introducing a number of improvements that will be incorporated into a future program.

Paper Nr: 16
Title:

Strategies of Multi-Step-ahead Forecasting for Blood Glucose Level using LSTM Neural Networks: A Comparative Study

Authors:

Touria El Idrissi, Ali Idri, Ilham Kadi and Zohra Bakkoury

Abstract: Predicting the blood glucose level (BGL) is crucial for self-management of Diabetes. In general, a BGL prediction is done based on the previous measurements of BGL, which can be taken either (manually) by using sticks or (automatically) by using continuous glucose monitoring (CGM) devices. To allow the diabetic patients to take appropriate actions, the BGL predictions should be done ahead of time; thus a multi-step ahead prediction is suitable. Therefore, many Multi-Step-ahead Forecasting (MSF) strategies have been developed and evaluated, and can be categorized in five types: Recursive, Direct, MIMO (for Multiple Input Multiple Output), DirMO (combining Direct and MIMO) and DirRec (combining Direct and Recursive). However, none of them is known to be the best strategy in all contexts. The present study aims at: 1) reviewing the MSF strategies, and 2) determining the best strategy to fit with a LSTM Neural Network model. Hence, we evaluated and compared in terms of two performance criteria: Root-Mean-Square Error (RMSE) and Mean Absolute Error (MAE), the five MSF strategies using a LSTM Neural Network with an horizon of 30 minutes. The results show that there is no strategy that significantly outperformed others when using the Wilcoxon statistical test. However, when using the Sum Ranking Differences method, MIMO is the best strategy for both RMSE and MAE criteria.

Paper Nr: 27
Title:

Med2Meta: Learning Representations of Medical Concepts with Meta-embeddings

Authors:

Shaika Chowdhury, Chenwei Zhang, Philip S. Yu and Yuan Luo

Abstract: Distributed representations of medical concepts have been used to support downstream clinical tasks recently. Electronic Health Records (EHR) capture different aspects of patients’ hospital encounters and serve as a rich source for augmenting clinical decision making by learning robust medical concept embeddings. However, the same medical concept can be recorded in different modalities (e.g., clinical notes, lab results) — with each capturing salient information unique to that modality — and a holistic representation calls for relevant feature ensemble from all information sources. We hypothesize that representations learned from heterogeneous data types would lead to performance enhancement on various clinical informatics and predictive modeling tasks. To this end, our proposed approach makes use of meta-embeddings, embeddings aggregated from learned embeddings. Firstly, modality-specific embeddings for each medical concept is learned with graph autoencoders. The ensemble of all the embeddings is then modeled as a meta-embedding learning problem to incorporate their correlating and complementary information through a joint reconstruction. Empirical results of our model on both quantitative and qualitative clinical evaluations have shown improvements over state-of-the-art embedding models, thus validating our hypothesis.

Paper Nr: 33
Title:

Impact of Threshold Values for Filter-based Univariate Feature Selection in Heart Disease Classification

Authors:

Houda Benhar, Ali Idri and Mohamed Hosni

Abstract: In the last decade, feature selection (FS), was one of the most investigated preprocessing tasks for heart disease prediction. Determining the optimal features which contribute more towards the diagnosis of heart disease can reduce the number of clinical tests needed to be taken by a patient, decrease the model cost, reduce the storage requirements and improve the comprehensibility of the induced model. In this study a comparison of three filter feature ranking methods was carried out. Feature ranking methods need to set a threshold (i.e. the percentage of the number of relevant features to be selected) in order to select the final subset of features. Thus, the aim of this study is to investigate if there is a threshold value which is an optimal choice for three different feature ranking methods and four classifiers used for heart disease classification in four heart disease datasets. The used feature ranking methods and selection thresholds resulted in optimal classification performance for one or more classifiers over small and large heart disease datasets. The size of the dataset takes an important role in the choice of the selection threshold.

Paper Nr: 34
Title:

Language Identification for Short Medical Texts

Authors:

Erick V. Godinez, Zoltán Szlávik, Selene B. Santamaría and Robert-Jan Sips

Abstract: Language identification remains a challenge for short texts originating from social media. Moreover, domain-specific terminology, which is frequent in the medical domain, may not change cross-linguistically, making language identification even more difficult. We conducted language identification on four datasets, two of them with general language, and two of them containing medical language. We evaluated the impact of two embedding representations and a set of linguistic features based on graphotactics. The proposed linguistic features reflect the graphotactics of the languages included in the test dataset. For classification, we implemented two algorithms: random forest and SVM. Our findings show that, when classifying general language, linguistic-based features perform close to the embedding representations of fastText and BERT. However, when classifying text with technical terms, the linguistic features outperform embedding representations. The combination of embeddings with linguistic features had a positive impact on the classification task under both settings. Therefore, our results suggest that these linguistic features could be applied for big and small datasets keeping the good performances in both general and medical languages. As future work, we want to test the linguistic features for a more significant set of languages.

Paper Nr: 49
Title:

A Patient’s Perspective on Decision-making for the Adoption of Digital Care Pathways

Authors:

Raja M. Abbas, Noel Carroll and Ita Richardson

Abstract: Healthcare Information Systems (HIS) are implemented to provide high-quality, patient-centred care. Yet, there is little evidence about the decision-making role patients play for the adoption of HIS nor what factors patients deem essential in the adoption of HIS. To guide healthcare practitioners in decision-making for the adoption of HIS, this study reports on the key factors which influence patients’ perception and use of HIS. Specifically, a qualitative study was conducted with 15 patients to understand the phenomenon of patient decision-making for the adoption of HIS. Our findings identify the concept of ‘Digital Care Pathways’ and indicate that there are four primary decision factors which influence the adoption of HIS: (i) trust; (ii) fear; (iii) ease of use; and (iv) accessibility. To synthesise the findings, we present the patients decision-making framework for digital care pathways as a first step to encapsulate the patients’ perspective of decision-making factors associated with adopting innovations for digital care pathways.

Paper Nr: 51
Title:

Data and Sessions Management in a Telepathology Platform

Authors:

Pedro Nunes, Rui Jesus, Rui Lebre and Carlos Costa

Abstract: Digital pathology refers to the acquisition, storage, and interpretation of pathological data gathered by scanners and displayed in a digital environment, using a distributed network system. This paper discusses the challenges and opportunities of a collaborative platform applied to digital pathology considering the advantages that it may carry on regarding education and training but also new paradigms of telepathology and telemedicine. Furthermore, it proposes the implementation of a secure collaborative platform that integrates a web pathology viewer with personal areas and virtual archives. The described approach introduces a modern collaborative concept into the digital pathology workflow supported by a customized medical imaging infrastructure where data management is ensured by an innovator DICOM standard multi-repository server. The solution was designed to serve distinct usage contexts, including telepathology and e-academy.

Paper Nr: 52
Title:

Status of Resources for Information Technology to Support Health Information Exchange in Resource-constrained Settings

Authors:

Andrew A. Egwar, Richard Ssekibuule and Josephine Nabukenya

Abstract: Various resources exist to support health information exchange (HIE). Both computerised and uncomputerized communication resources continue to be used in resource-constrained environments, like the Uganda health system to support HIE. Despite the rapid shift to the digital health environment, the resource capabilities of health systems in LMICs to support robust HIE is unknown. This study surveyed the status of resources for ICT to support ehealth communication in a resource-constrained setting. The study was conducted in three districts, representing the urban, peri-urban and rural settings of Uganda. The qualitative data collected was analysed with QSR NVivo 10. Results show major resource challenges including financial constraints, funders restrictions, human resource limitations, isolated computer systems, lack of support from management, legacy/outdated systems, intermittent/limited network bandwidth, limited hardware, misuse/poor maintenance of the available hardware, and power outages among others. In addition, results show a great disparity in their distribution across the healthcare sector. Therefore, we argue that much improvement is needed if the benefits of ehealth are to be attained in LMICs. Recommendations include specifying minimum resources for ICT required to support HIE, supervising implementation and monitoring compliance to the standards, establish a mechanism for periodic review of the minimum standards, and finally, align ICT funding within the mainstream funding for healthcare services. It should uniformly apply across the board (i.e., facilities located in urban, peri-urban and urban) for the full benefits of ICT in health to be achieved in LMICs.

Paper Nr: 56
Title:

Developing a Machine Learning Model for Predicting Postnatal Growth in Very Low Birth Weight Infants

Authors:

Andrea Seveso, Valentina Bozzetti, Paolo Tagliabue, Maria L. Ventura and Federico Cabitza

Abstract: Objective of the work is the development of prognostic machine learning models that predict qualitative and quantitative measures of postnatal growth in very low birth weight preterm infants. Observational retrospective data about 964 infants at risk are retrieved from “Fondazione Monza e Brianza per il bambino e la mamma“’s electronic medical record. Both prenatal (gestational, socioeconomic, etc.) and perinatal (nutritional, respiratory assistance, drug prescription and daily growth) data up to a week after birth are the features included. Model’s performances are compared to previous literature and human performance, showing a substantial improvement (in e.g., best regression MAE=0.49, best classification AUC=0.94).

Paper Nr: 61
Title:

Applying Agile Principles to Collaborative Healthcare Teams

Authors:

Rubina Lakhani, Benjamin Eze and Liam Peyton

Abstract: In this paper, we demonstrate how agile principles can be applied to collaborative healthcare teams. We provide a generic Agile Healthcare Process, and two associated artifacts, the Agile Treatment Plan, and the Agile Dashboard using a theoretical Attention Deficit Hyperactivity Disorder case study. The paper describes these in detail and shows how healthcare teams can measure the success of their collaboration through actionable metrics. Our hypothesis is that providing a process in which collaboration factors are identified and associated with specific performance metrics that can be collected and analyzed, can improve coordination of collaborative healthcare teams. We demonstrate how agile methodology can be applied to manage the treatment of chronic conditions such as ADHD. Our approach anchors around the Agile Treatment Plan and the Agile Dashboard. We show how the KPIs associated with these artifacts can be used to quantify healthcare team collaboration and performance.

Paper Nr: 70
Title:

A Process Reference Model for Enhanced Medication Management

Authors:

Bonnie Urquhart and Waqar Haque

Abstract: A comprehensive management approach to improve quality and safety of medication management within a multisite health care organization is explored. The process-oriented approach integrates Business Ontology, Business Architecture and Business Process Management to develop a reference model for medication management. The model includes one hundred and sixty-four individual processes categorized in four process areas and twenty-five process groups. The business artefacts and methodology used in the development of this reference model are also presented. These artefacts were created and validated through workshops and working group meetings. The developed methodology could be used to create a similar process architecture in other organizations and service areas.

Paper Nr: 71
Title:

Are the Healthcare Institutions Ready to Comply with Data Traceability Required by GDPR? A Case Study in a Portuguese Healthcare Organization

Authors:

Cátia Santos-Pereira, Alexandre B. Augusto, José Castanheira, Tiago Morais and Ricardo Correia

Abstract: GDPR introduces a new concept: ”Data protection by design and per default” for new software development however legacy systems will also have to adapt in order to comply. This creates great pressure on health care institutions, namely hospitals, and software producers to provide data protections and traceability mechanisms for their current and legacy systems. The aim of this work is to understand the maturity level of a Portuguese Healthcare Organization in their audit records to comply with GDPR article 30 and 32 since healthcare organization operate in a daily-basis with personal data. This study was performed with the partnership of a public Portuguese healthcare organization and were organized into three main phases: (1) data collection of all information systems that operate with personal data; (2) interviews with IT professionals in order to retrieve the necessary knowledge for each information system and (3) analysis of the collected data and its conclusions. This study helped to identify a need inside this organization and to determine a follow-up plan to overpass this challenge. However it also identified some constrains like financial budget, legacy systems, small team of IT professionals in the organization and difficulties in establish communication with information system providers.

Paper Nr: 73
Title:

Analyzing Privacy Practices of Existing mHealth Apps

Authors:

Aarathi Prasad, Matthew Clark, Ha L. Nguyen, Ruben Ruiz and Emily Xiao

Abstract: Given students’ reliance on smartphones and the popularity of mobile health apps, care should be taken to protect students’ sensitive health information; one of the major potential risks of the disclosure of this data could be discrimination by insurance companies and employers. We conducted an exploratory study of 197 existing smartphone apps, which included 98 mobile health apps, to study their data collection, usage, sharing, storage and deletion practices. We present our findings from the analysis of privacy policies and permission requests of mHealth apps, and propose the need for a usable health data dashboard for users to better understand and control how their health data is collected, used, shared and deleted.

Paper Nr: 76
Title:

Activity Scores of Older Adults based on Inertial Measurement Unit Data in Everyday Life

Authors:

Sandra Hellmers, Lianying Peng, Sandra Lau, Rebecca Diekmann, Lena Elgert, Jürgen M. Bauer, Andreas Hein and Sebastian Fudickar

Abstract: The trend of an ageing population is becoming more and more obvious. Staying healthy in old age is an important social issue. Thereby, physical activity is essential for the preservation of physical function. We developed an algorithm for determining the activity level of seniors in everyday life. The proposed algorithm is based on machine learning activity detection using inertial measurement unit data. A series of activity scores is obtained by executing the algorithm from data on the type of activity, total activity time and activity intensity. To evaluate the performance of the proposed algorithm, a study with 251 participants aged above 70 (75.41 ± 3.88) years was conducted and the correlation between individual activity scores and clinical mobility assessments was determined. Results showed a relation between the Six Minute Walking Test and the total score in terms of activity level as well as the walk score. Additionally, the MVPA- and walk-score show a clear trend regarding the frailty status of the participants. Therefore, these scores are indicators of the physical function and hence validate the utility of the developed algorithm.

Paper Nr: 83
Title:

Smart Community Health: A Comprehensive Community Resource Recommendation Platform

Authors:

Mehdi Mekni and David Haynes

Abstract: Health disparities and inequities are explained by the conditions of places where people live, learn, work and play. In fact, the health of an individual is partially related to access and quality of health care and mainly associated to his behaviours, socioeconomic conditions and other community related factors that are often challenging to address by health care organizations. To meet the need for information about local social services organizations and the ability to offer resource referrals, a number of platforms have been proposed that provide electronic social resource directories and facilitate referrals to social service agencies. However, these platforms show limitations with regards to their dependancy to health care organizations, application portability, service availability, and user engaging interactions such as tracking, monitoring and notification. Moreover, existing social resource referral platforms suffer from a fragmentation of services and a disconnection between individuals in need and service providers. In this paper, we introduce Smart Community Health (SCH), a novel independent platform that prioritizes connecting people in need with local community resources. SCH is a full-service, end-to-end community service provider recommendation platform designed to help address pressing social, environmental, and health needs within our communities. The platform is composed of a mobile application for individuals looking for services and a web application dashboard for the management of community service providers and health care organizations.

Paper Nr: 92
Title:

Methods for Preclinical Validation of Software as a Medical Device

Authors:

Alice Ravizza, Federico Sternini, Alice Giannini and Filippo Molinari

Abstract: Software as a medical device is subject to dedicated regulatory requirements before it can be used on human beings. The certification process in Europe requires that sufficient data on clinical benefits are available before the device is CE marked. This position paper describes our proposal of a risk-based approach to technical and preclinical validation of software as medical devices. This approach ensures that all technical solutions for safety are implemented in the software and that all information for safe use is consistent before the software can be made available to patients. This approach is compliant to the main international standards ISO 13485 on quality systems and ISO 14971 on risk management and therefore ensures regulatory compliance as well as patient protection. This integrated approach allows the designers of the software to integrate regulatory and safety testing in the technical testing of the candidate release version of the device. This approach ensures patient safety and regulatory compliance at the same time as technical functionality.

Paper Nr: 97
Title:

Individualized Computer-based Training for Elderly in Nursing Homes: A Pilot Study

Authors:

Katja Orlowski, Gina M. Gräfe, Laura Tetzlaff, Thomas Schrader and Eberhard Beck

Abstract: In older ages, the people are affected by limitations referring to physical and cognitive functions of the body. These limitations can lead to falls, which can be prevented by different types of physical training. Some studies showed that different kinds of physical activity have a positive effect on the equilibrium as well as on cognitive function. During a project an individualized computer-based training was developed. The developed application was examined during a pilot study in a local nursing home. The results indicate that the training intervention based on the computer-based training has a positive effect on different parameters (balance, TUG test). The limitation of the pilot study is the small sample size, which is additionally reduced due to dropouts. In further studies the effect of a balance training with the computer-based training will be done in other nursing homes.

Paper Nr: 98
Title:

Adopting the Mediterranean Diet Score in a Diet Management System

Authors:

Luca Anselma, Mirko Di Lascio, Antonio Lieto and Alessandro Mazzei

Abstract: In this work we want to study the possibility to integrate a Mediterranean Diet Score into an existing diet management architecture. The main goal is to integrate the quantitative constraints on macronutrients with the qualitative constraints encoded in the Mediterranean Diet. This paper presents some preliminary results on this roadmap.

Paper Nr: 99
Title:

SalivaPrint as a Non-invasive Diagnostic Tool

Authors:

Eduardo Esteves, Igor Cruz, Ana C. Esteves, Marlene Barros and Nuno Rosa

Abstract: Currently, the molecular diagnosis is based on the quantification of RNA, proteins and metabolites because they present changes in their quantity related to clinical situations. The same molecules are not generally suitable for early diagnosis or to follow clinical evolution, making necessary strategies to evaluate the complete molecular scenario. There are already experimental strategies that allow the determination of total protein profiles from saliva samples (the SalivaPrint). The goal of this work is to identify a profile of saliva proteins (similar to a fingerprint) and, using computational methods, identify how this profiles changes with age and gender. So far it has been possible to collect 79 samples as well as the metadata associated with each sample using an electronic questionnaire developed by us. A total protein profile was obtained and their association with gender was verified using statistical methods. Currently we are developing the Python scripts for automatic data acquiring and normalization. Total protein profiles annotation on a database (SalivaPrintDB) and their integration with the factors that affects them using machine learning strategies can empower the use of the approach proposed on this work as a tool for monitoring the individual's health status.

Paper Nr: 105
Title:

Differences in Brain Activity of Skilled and Novice Nurses during Blood Collection

Authors:

Naoki Taira, Yukie Majima, Seiko Masuda, Tsuneo Kawano, Masanori Akiyoshi, Kenji Adachi, Kazuma Mihara and Ryoma Namba

Abstract: Nursing skills are highly implicit. No effective method has been established for passing them on to the next generation of nursing workers. This research represented an attempt to formalize the skills of skilled nurses. To evaluate human mental state and emotions objectively, we clarified differences in brain activity between skilled and novice nurses at the time of collecting blood from a patient. As a result, many skilled nurses tried to use their own knowledge for the blood vessels they faced for the first time, irrespective of the blood collection success or failure. However, some novice nurses face subsequent blood collection without reflecting on the clear factors leading to success or failure.

Paper Nr: 108
Title:

Process Mining of Disease Trajectories: A Feasibility Study

Authors:

Guntur P. Kusuma, Samantha Sykes, Ciarán McInerney and Owen Johnson

Abstract: Modelling patient disease trajectories from evidence in electronic health records could help clinicians and medical researchers develop a better understanding of the progression of diseases within target populations. Process mining provides a set of well-established tools and techniques that have been used to mine electronic health record data to understand healthcare care pathways. In this paper we explore the feasibility for using a process mining methodology and toolset to automate the identification of disease trajectory models. We created synthetic electronic health record data based on a published disease trajectory model and developed a series of event log transformations to reproduce the disease trajectory model using standard process mining tools. Our approach will make it easier to produce disease trajectory models from routine health data.

Paper Nr: 114
Title:

Informatics as Support for Changes in Health Policy: A Case in Obstetrics

Authors:

Giovana J. Gelatti, Pedro P. Rodrigues and Ricardo João C. Correia

Abstract: Introduction: In 2015 the Directorate-General for Health of Portugal published new standards (DGS 001/2015) for the registration of cesarean section indicators. The existing scenario was the lack of data, influencing the quality of indicators and analyses on them. The use of a single computer tool was encouraged to register and compare indicators between hospitals with special attention to the Robson Classification as it employs basic information of pregnancy to classify all deliveries in 10 groups. The selected tool was Obscare software. Aim: Describe the scenario on data quality by analyzing the completeness of obstetric records from 2016 to 2018 of the variables used in Robson’s classification collected by the Obscare tool. Methods: The completeness is evaluated using a number of missing values. The lower the completeness, the higher the number of missing values. Also, we perform the imputation of data based on basic concepts and analyzed the participation of this data in the indication of the type of delivery to be performed according to classification suggested by DGS 001/2015. Results: From 2016 to 2018, 5922 number of pregnancies resulted in 5922 of Robson Classifications. The variables with lower completeness were related to previous cesarean section (77%) and previous pregnancies (43%). After imputation, it fell to 3.9% and 0.56%, respectively causing 4.6% of discarded data from the total. Discussion: There is a significant amount of missing data in basic variables used to study the classification of delivery type. We believe that encouraging data completion with the possibility of comparing data between hospitals should be a priority in the health area.

Paper Nr: 119
Title:

Human Factors Standards and the Hard Human Factor Problems: Observations on Medical Usability Standards

Authors:

Lorenzo Strigini and Marwa Gadala

Abstract: With increasing variety and sophistication of computer-based medical devices, and more diverse users and use environments, usability is essential, especially to ensure safety. Usability standards and guidelines play an important role. We reviewed several, focusing on the IEC 62366 and 60601 sets. It is plausible that these standards have reduced risks for patients, but we raise concerns regarding: (1) complex design trade-offs that are not addressed, (2) a focus on user interface design (e.g., making alarms audible) to the detriment of other human factors (e.g., ensuring users actually act upon alarms they hear), and (3) some definitions and scope restrictions that may create “blind spots”. We highlight potential related risks, e.g. that clear directives on “easier to understand” risks, though useful, may preclude mitigating other, more “difficult” ones; but ask to what extent these negative effects can be avoided by standard writers, given objective constraints. Our critique is motivated by current research and incident reports, and considers standards from other domains and countries. It is meant to highlight problems, relevant to designers, standards committees, and human factors researchers, and to trigger discussion about the potential and limits of standards.

Paper Nr: 8
Title:

Examination of Interpersonal Attachment with the Help of a Digital Tablet Application: A Proof of Concept Study

Authors:

Sebastian Unger, Cony Theis and Thomas Ostermann

Abstract: At present, interpersonal attachment has a subordinate role in the field of healthcare, but recent research results assume this as an important parameter, especially in prevention and mental health. Our aim was to develop a digital application that extends the previous approaches with a measurement over a specific time interval. Designed specifically for Windows-based tablets, this application performs a drawing test while capturing the transitions of two mental states, transmitted by the users. The results were collected over a period of three minutes, allowing the application itself, along with the SiZer analysis, to determine how closely the participants were mentally connected. The tablet application has shown its first usefulness to enhance the healthcare, but further investigations are strongly recommended. In addition, its ease to use allows an uncomplicated integration into similar areas.

Paper Nr: 12
Title:

Challenges and Opportunities for Caregiving through Information and Communication Technology

Authors:

Jane Moloney, Raja M. Abbas, Sarah Beecham, Bilal Ahmad and Ita Richardson

Abstract: The increased susceptibility of world’s population to diseases augmented with the decrease in the healthcare workforce leads to over-reliance on caregivers. This increased burden on caregivers adversely impacts their quality of life. However, information and communication technology (ICT) has the potential to facilitate caregiving. Therefore, the objective of this study is to investigate the opportunities and challenges for caregiving through ICT, the development of a prototype to support caregivers better monitor their care recipients (known as clients) and the evaluation of that prototype. A qualitative study with 10 caregivers was conducted to address the research questions from which data was coded and analysed. Using this data, a web-based prototype was developed and evaluated by 5 caregivers and 5 technology experts. The instruments used were interviews and focus groups. The results revealed eight categories for improving care identified by caregivers.

Paper Nr: 18
Title:

Interrelations between Drug Prescriptions and Diagnoses for SHI Diabetes Patients using Graph Theoretic Methods and a Markov Model

Authors:

Reinhard Schuster, Marc Heidbreder, Timo Emcke and Martin Schuster

Abstract: We analyze large data sets of diabetes patients in order to get new insights into the dependencies between drug groups and diagnoses using age, polypharmacy and multimorbidity as covariates. Diagnostic data using the ICD-10 classification are available with the resolution of quarters. For drugs the exact day of prescription is available. The analysis uses all co-medication and all diagnoses of all physicians a patient has consulted within a quarter and is thereby wider than the point of view related to a special physician. The communication between physicians may be confounded by information deficits due to informal self-diagnostics by the patients. Differently specialized physicians may apply different guidelines which point to specific diseases. Interactions between different drugs and different therapy schemes may lead to new diseases for multimorbid patients. Large data sets create opportunities to detect such interactions. We use a graph theoretic approach with drug groups as nodes. Using a diagnose vector edges are given by therapeutic neighborhood using the Manhattan distance. A graph clustering determines drug groups for similarly sick patients which contains indirectly age and multimorbidity. This can explain cost effects due to the degree of sickness. The graph clustering uses the modularity method. The underlying algorithm leads to an integer linear program (ILP) which is in general NP-hard. For the calculations we use Mathematica from Wolfram Research in combination with a python program using CPLEX from IBM. Drug innovations may lead to changes in drug therapy. Therefore we compare the steady state solution of the related Markov model with the status quo of drug prescription.

Paper Nr: 19
Title:

IT-structures and Algorithms for Quality Assurance in the Health Insurance Medical Advisory Service Institutions in Germany

Authors:

Vera Ries, Klaus-Peter Thiele, Martin Schuster and Reinhard Schuster

Abstract: The 15 Regional Medical Advisory Service Institutions of the Statutory Health Insurance in Germany (MDK) create medical expertises (sozialmedizinische Gutachten) on behalf of the Statutory Health Insurance Funds in the fields of inpatient and outpatient treatment, incapacity of work and other fields. The process of internal quality assurance within the local advisory service institutions as well as between them plays an increasing role and got new impulses by actual national legislation. The assessment process was established in 2004 and covers only one single indication: long-term care. It consists in a paper-based procedure focusing on a manual distribution process performed by staff of the central quality assurance bureau. We will present organizational concepts of the new standardized regional and nationwide peer review process that will cover the multitude of all existing indications of health care. It is completely digitalized using mathematical IT-based procedures not only for randomized sampling, but as well to achieve an equal distribution of the medical expertises to be assessed by the peer Medical Advisory Service Institutions. These peer reviews are supposed to be distributed among the institutions depending on occasion groups and further subtypes of medical expertises, posing a constraint satisfaction problem that needs to be solved. Therefore we discuss models that address this kind of problem type and present possible solutions for the concrete distribution problem mentioned above. We further present our technical framework that will support the workflow needed for peer review distribution, data collection and final result analysis. The 15 regional medical service institutions with different IT-system have to be connected, while data protection concerns have to be taken into account. Finally, the statistical distribution of the review results is analyzed.

Paper Nr: 21
Title:

Operationalizing Healthcare Big Data in the Electronic Health Records using a Heatmap Visualization Technique

Authors:

Don Roosan, Mazharul Karim, Jay Chok and Moom R. Roosan

Abstract: Background: The majority of the electronic health record (EHR) contains a wealth of information, including unstructured notes. Healthcare professionals may be missing substantial portions of essential diagnostic and treatment information by not focusing on unstructured texts. The objective of this study is to present progress notes data using heatmap visualization. Methods: In this study, the research team used the unstructured text from the progress notes of deidentified patient data. The research team conducted qualitative content-coding based on the clinical complexity model and developed a heatmap based on the processed frequency data. Result: The researchers developed a color-coded heatmap focusing on the severity and acuity of patients’ status accumulated through multiple previous patient’s visits. Conclusions: Future research into creating an automated process to generate the heatmap from an unstructured dataset can open up opportunities to operationalize big data in healthcare.

Paper Nr: 30
Title:

Architecture of a Learning Surveillance System for Malaria Elimination in India

Authors:

S D Sreeganga, Susanna G. Mitra and Arkalgud Ramaprasad

Abstract: Surveillance is critical for malaria elimination. Malaria transmission takes place in a dynamic and complex environment. The key goal in developing a malaria surveillance system is to ensure that it is robust, systematic, and effective for improving data availability for decision-making. We present a unified framework for envisioning malaria surveillance informatics as an ontology-based feedback system. The framework presented is a solution for the current fragmented and linear surveillance processes for malaria case management. It encapsulates a comprehensive natural language enumeration of the requirements of the cyberenvironment, structured into 5 dimensions - timing, surveillance process, information surveyed, malaria management, and stakeholder, with each of them articulated as a taxonomy of its constituent elements. The elements are combined to form natural language statements of the cyberenvironment requirement. The information generation through the semiotic cycle provides real-time sense and response capability for timely and targeted interventions. The response mechanism creates both positively and negatively reinforcing feedback-based learning processes at multiple levels. Such a system enables data interoperability for capturing malaria incidence, discover epidemiological clusters, and predict propagation dynamics. On a larger scale, the integrative framework enables data harmonization, analytics, and visualization towards effective management and knowledge generation on disease surveillance.

Paper Nr: 31
Title:

Towards an On-line Handwriting Recognition Interface for Health Service Providers using Electronic Medical Records

Authors:

Viktor Mikhael M. Dela Cruz, Christian E. Pulmano and Ma. Regina Justina E. Estuar

Abstract: The 2019 Universal Health Care Act in the Philippines has allowed healthcare service providers to have a second look at using electronic medical records (EMRs) in their practice with tools that enable servicing the poorest of the poor and coursing payments via EMR. A review of first world country narratives, however, show evidence of the substandard usability of EMRs. Physician work is impeded as almost two-thirds of consultation time is spent documenting on an EMR instead conversing with patients face-to-face. This paper describes a handwriting recognition interface for EMR data entry that is user-friendly and is unobstructive to the patient-physician relationship. An initial prototype tested by medical students showed a handwriting recognition accuracy of 34% while a second testing by health service providers showed a handwriting recognition accuracy of 42%. Findings show that recognition is challenged by specialized words and accidental markings which cause extra spaces and extra symbols. Additional features to the system as well as possible augmentations to improve accuracy and efficiency through ontology, machine learning, and AI are also roadmapped.

Paper Nr: 35
Title:

Creating Patient Decision Aid Tools

Authors:

Andrea Corradini, Constantin A. Gheoghiasa and Jesper Nordentoft

Abstract: This paper reports on the creation of a web application that facilitates the development and implementation of patient decision aid tools. We propose a software prototype model that allows medical personnel to easily and rapidly create digital prototypes of patient decision aid tools independently on the medical condition. Our application can be used as an online framework and is being tested by healthcare professionals.

Paper Nr: 36
Title:

VisualMLTCGA: An Easy-to-Use Web Tool for the Visualization, Processing and Classification of Clinical and Genomic TCGA Data

Authors:

Alba Garin-Muga, Aurora M. Sucre, Jordi Torres and Jon Kerexeta

Abstract: The Cancer Genome Atlas (TCGA) is a collection of freely available data of several human cancer types. TCGA contains over 2.5 petabytes of data, which includes, among others, clinical and genomic data. However, the visualization of such data is cumbersome and tiring for non-expert users. VisualMLTCGA is an intuitive and easy-to-use web tool that allows the automatic download and visualization of TCGA data and the processing of genomic data using GATK. Additionally, the tool allows to create comprehensive decision trees (DT) for prediction of outcomes from clinical and genomic TCGA data and other external datasets. VisualMLTCGA offers a simple web tool to download, process and visualize TCGA data, suitable for researchers and clinicians without any bioinformatics background.

Paper Nr: 39
Title:

Multimodal Fusion Strategies for Outcome Prediction in Stroke

Authors:

Esra Zihni, Vince Madai, Ahmed Khalil, Ivana Galinovic, Jochen Fiebach, John D. Kelleher, Dietmar Frey and Michelle Livne

Abstract: Data driven methods are increasingly being adopted in the medical domain for clinical predictive modeling. Prediction of stroke outcome using machine learning could provide a decision support system for physicians to assist them in patient-oriented diagnosis and treatment. While patient-specific clinical parameters play an important role in outcome prediction, a multimodal fusion approach that integrates neuroimaging with clinical data has the potential to improve accuracy. This paper addresses two research questions: (a) does multimodal fusion aid in the prediction of stroke outcome, and (b) what fusion strategy is more suitable for the task at hand. The baselines for our experimental work are two unimodal neural architectures: a 3D Convolutional Neural Network for processing neuroimaging data, and a Multilayer Perceptron for processing clinical data. Using these unimodal architectures as building blocks we propose two feature-level multimodal fusion strategies: 1) extracted features, where the unimodal architectures are trained separately and then fused, and 2) end-to-end, where the unimodal architectures are trained together. We show that integration of neuroimaging information with clinical metadata can potentially improve stroke outcome prediction. Additionally, experimental results indicate that the end-to-end fusion approach proves to be more robust.

Paper Nr: 40
Title:

Music Recommendation System for Old People with Dementia and Other Age-related Conditions

Authors:

Miriam Allalouf, Avi Cohen, Lea C. Sabban, Ayelet Dassa, Sagi Marciano and Stella M. Beris

Abstract: The worldwide increase in life expectancy can be accompanied by age-related degenerative conditions such as dementia. Dementia poses significant challenges for which music is a beneficial non-pharmacological intervention. Based on research and clinical expertise we developed a web-based system, termed Tamaringa, that builds and displays customized playlists. The recommendation mechanism incorporates an old person's age, birthplace, and popular songs from their youth. That particular range is known as being most accessible to seniors in terms of memory. Although there are a lot of repositories containing metadata and information about music, there is no single repository that addresses all our requirements in terms of specific metadata, range query application programming interfaces (API) capability and popularity information. This study explores the APIs of several repositories in order to populate our internal database with suitable songs that are required for accurate personalized recommendation. A preliminary promising pilot enabled twenty-four residents in an assisted living facility in Israel to engage and enjoy the music recommendation system. Personalized playlists were created using the system; the medical staff reports were positive. Further research will help to develop our system and eventually to integrate its use both in assisted living facilities and at home.

Paper Nr: 46
Title:

Towards an Ambient Support System for Continence Management in Nursing Homes: An Exploratory Study

Authors:

Hannelore Strauven, Ine D’Haeseleer, Kristof T’Jonck, Hans Hallez, Vero V. Abeele, Pieter Crombez and Bart Vanrumste

Abstract: Time consuming and costly, continence care management has become one of the main care demands in nursing homes with potential inadequacy negatively impacting residents’ quality of life. While engineering efforts in this area are increasing, these mainly focus on wearable innovations. To support continence care in nursing homes in an unobtrusive manner, we developed an ambient sensor system to continuously monitor incontinence events, e.g., saturated incontinence materials or leakages. In an exploratory study in two nursing homes, we evaluated an early prototype of the sensor system and built annotated data sets. Implemented annotation devices included a smart sensor mat, a toilet timing predicting device, and manual data entry of continence care by care personnel. From our analysis of the preliminary study results based on the first two residents, we learned how challenging the ambient monitoring and annotation of incontinence events is. On the basis of the outcomes, we provide suggestions for further research of ambient sensor systems supporting continence care.

Paper Nr: 53
Title:

Interpretation of Patients’ Location Data to Support the Application of Process Mining Notations

Authors:

Sina N. Araghi, Franck Fontanili, Elyes Lamine, Nicolas Salatge and Frederick Benaben

Abstract: The application of indoor localization and process mining emerges as an intriguing tool for the researchers to address the structural issues related to the patient pathways inside healthcare organizations. However, there is a major gap in the literature. This is related to the lack of enough attention to the interpretation of location data. Therefore, as a contribution, this article presents the DIAG meta-model and relevant location data interpretation rules. This model-driven approach has been realized in the context of the R-IOSUITE application and it supports the further analyses by the process mining methods.

Paper Nr: 55
Title:

Blue Light and Melanopsin Contribution to the Pupil Constriction in the Blind-spot, Parafovea and Periphery

Authors:

Tim Schilling, Mojtaba Soltanlou, Yeshwanth Seshadri, Hans-Christoph Nuerk and Hamed Bahmani

Abstract: Retinal photoreceptors modulate the pupil diameter to regulate retinal illumination. At early stage the pupil-response is formed by intrinsically-photosensitive-Retinal-Ganglion-Cells (ipRGCs) expressing melanopsin, activated by blue light. ipRGCs’ axons pass through the optic nerve head, corresponding to the blind-spot. No photoreceptors except melanopsin appear to exist in the blind-spot. Contributions of melanopsin to pupil constriction in absence of classical photoreceptors in the blind-spot is not fully understood. We investigated how blue light in the blind-spot changes melanopsin-pupil-response compared to parafovea and periphery. The Post-Illumination-Pupil-Response (PIPR) amplitude reflecting melanopsin was analyzed for standardized time windows (1s<1.7s, 1s>1.8s and 2–6s) and expressed as pupillary-change. Bayesian analysis showed a BF>3 that PIPR>1.8s for blind-spot and periphery is not different. At times 2s–6s, a t-test comparison in the blind-spot condition showed a significantly larger PIPR to blue compared to red light, confirming a melanopsin-pupil-response in the blind-spot. Taken together, equivalent stimulation in the blind-spot and periphery revealed comparable PIPR, although there are no rods and cones in the blind-spot. In absence of classical photoreceptors in the blind-spot, melanopsin seems to be responsible for pupil constriction in similar manner as in the periphery, which supports the presence of melanopsin on the axons of ipRGCs.

Paper Nr: 57
Title:

AdoBPRIM: Towards a New Healthcare Risk-aware Business Process Management Tool

Authors:

Rafika Thabet, Amine Boufaied, Elyes Lamine, Dominik Bork, Ouajdi Korbaa and Hervé Pingaud

Abstract: Performing risk management in healthcare facilities is particularly difficult due to the highly dynamic, complex, and multi-disciplinary nature of healthcare processes like the Medication Use Process (MUP). Risk-aware Business Process Management (R-BPM) is a promising approach to obtain a better understanding of such processes by identifying and analyzing corresponding risks. However, not all R-BPM approaches perform well in capturing the complexity of clinical processes. In this work, we introduce a new R-BPM framework called BPRIM that allows the identification and the analysis of medication error risks related to the complex medication use process. BPRIM is implemented using the ADOxx meta-modelling platform and then tested in a real case study. The tool is specific to the case study, but the framework can be used also in other healthcare processes.

Paper Nr: 63
Title:

A Framework for System-level Health Data Sharing

Authors:

Mana Azarm, Craig Kuziemsky and Liam Peyton

Abstract: Circle of care is the term that has been used to provide context for health data sharing that is allowed by privacy regulation that occurs when a diverse team is collaborating to provide care to a patient. We introduce the concept of system-level health data sharing to capture the totality of health data that exists for a patient in a healthcare system across multiple health care organizations. MyPHR is a system-level health data-sharing framework that guides any healthcare system to set up interoperable, patient-centred health data sharing. We briefly introduce the components of MyPHR framework and then discuss its evaluation by a panel of experts who reviewed a demonstration walkthrough of the interfaces and data sharing that the framework supports.

Paper Nr: 64
Title:

A Rule-based Content Management Framework for Effective Development of Intelligent Mobile Apps in Healthcare

Authors:

Mohammad Raahemi, Benjamin Eze, Cléo Mavriplis and Liam Peyton

Abstract: The number of published healthcare articles is increasing dramatically every year, making it difficult for physicians and patients to stay current with the latest information related to healthcare. One possible approach to improving the ability of physicians and patients to stay current with the latest trends in healthcare is through the use of mobile applications. The challenge to this approach is the lack of a content management framework that allows medical experts to continuously integrate new knowledge and content into the design of easy to use software applications for patients and other healthcare personnel. This paper introduces the CANBeWell mobile application, a rule-based content management application for collecting and aggregating important medical data from medical experts, and disseminating this data to patients and other clinicians using a context-aware mobile app in support of preventive healthcare.

Paper Nr: 65
Title:

Dietitians and Nutritionists Behaviour on Social Media: A Scoping Literature Review

Authors:

Inga Saboia, Ana P. Almeida, Pedro Sousa and Cláudia Pernencar

Abstract: At its present state, Social Media (SM) is an important stage to promote user participation, acting as an open space for the discussion of a multitude of fields, one of which being health. Professionals, like Registered Nutritionists and Dietitians (RNDs), whose access to media was traditionally more restricted, are also more engaged in this new context, creating a new scenario. To better understand how is this group of professionals using social media to communicate with their audiences is the main objective of this study. To approach this topic, a mapping was conducted, followed by a presentation of the summary of the evidence discovered: RDNs demographic and professional profile; their most used social media tools; the reasons why they use social media; their common behaviours and attitudes, as well as a review of the gaps and shortcomings in the literature. A literature review, using a structured approach was also conducted. 2877 works were screened, but only 8 were associated with answers. Of these 8, there were 2 studies that partially presented a quantitative analysis. Results show lacks in consolidated studies that can be used to support the creation of knowledge in this field. This lead to conclude that research about social media usage by nutritionist, at present, remains in a nascent stage and requires further studies.

Paper Nr: 69
Title:

Illegitimate HIS Access by Healthcare Professionals Detection System Applying an Audit Trail-based Model

Authors:

Liliana Sá-Correia, Manuel E. Correia and Ricardo Cruz-Correia

Abstract: Complex data management on healthcare institutions makes very hard to identify illegitimate accesses which is a serious issue. We propose to develop a system to detect accesses with suspicious behavior for further investigation. We modeled use cases (UC) and sequence diagrams (SD) showing the data flow between users and systems. The algorithms represented by activity diagrams apply rules based on professionals’ routines, use data from an audit trail (AT) and classify accesses as suspicious or normal. The algorithms were evaluated between 23rd and 31st July 2019. The results were analyzed using absolute and relative frequencies and dispersion measures. Access classification was in accordance to rules applied. “Check time of activity” UC had 64,78% of suspicious classifications, being 55% of activity period shorter and 9,78% longer than expected, “Check days of activity” presented 2,27% of suspicious access and “EHR read access” 79%, the highest percentage of suspicious accesses. The results show the first picture of HIS accesses. Deeper analysis to evaluate algorithms sensibility and specificity should be done. Lack of more detailed information about professionals’ routines and systems, and low quality of systems logs are some limitations. Although we believe this is an important step in this field.

Paper Nr: 75
Title:

A Case Study of Conformance Checking for Diagnosis and Treatment of Ischemic Stroke based on Clinical Guidelines

Authors:

Haifeng Xu, Jianfei Pang, Xi Yang, Jinghui Yu, Huajian Mao and Dongsheng Zhao

Abstract: Clinical guidelines provide the best practices for therapy of patients with ischemic stroke, so it is necessary to check the actual implementation of clinical guidelines. Based on the analysis of clinical guidelines, we abstracted the relationship of medical events for patients with ischemic stroke using declarative process mining framework, and constructed the de jure model manually. To test our method, a real life data set from hospital is checked and analyzed with the handmade model of medical process. The results show that the manually generated model can recognize abnormal activities with flexibility, and the approach proposed in this paper could be applied to actual business environments.

Paper Nr: 78
Title:

Data Collection via Wearable Medical Devices for Mobile Health

Authors:

Vincenza Carchiolo, Alessandro Longheu, Simone Tinella, Salvo Ferrara and Nicolò Savalli

Abstract: The prevention and early detection of illness symptoms is becoming more and more essentials in a world where the improvements in healthcare extends life expectancy. New technologies led to new paradigma as e-health, m-health, smart-health and pervasive health. Wearable networked devices for real-time and self-health monitoring represent an effective approach that fulfil prevention goal at the same time keeping costs under control. In this work, we present a Wearable Health Monitoring Systems (WHMS) capable of collecting, digitizing, connecting to a wearable medical device via Bluetooth, and measuring various physiological parameters of patients in particular suffering from heart disease. System’s architecture, requirements, adopted technologies and implementation issues are presented and discussed, showing its effectiveness in healthcare support.

Paper Nr: 79
Title:

Sensor-based Solutions for Mental Healthcare: A Systematic Literature Review

Authors:

Nidal Drissi, Sofia Ouhbi, José A. García-Berná, Mohammed J. Idrissi and Mounir Ghogho

Abstract: Mental well-being is a crucial aspect of the person’s general health, compromised mental health impairs the person’s functioning, decreases the quality of life, and limits the person’s contribution to society. The mental health industry is still facing some barriers to healthcare delivery such as costs, mental health illiteracy, and stigma. Incorporating technological interventions in the treatment and the diagnosis processes might help overcome these barriers. Sensors are devices that have been used for healthcare since the 1990s and have been incorporated into mental healthcare in different forms. In this study, we conducted a systematic literature review to identify and analyze sensor-based solutions for mental healthcare. 12 studies were identified and analyzed. The majority of the selected studies presented methods and models and were empirically evaluated and showed promising accuracy results. Different types of sensors were used to collect different types of data about the patient such as physical and behavioral information. The selected studies mainly addressed the use of sensors for common mental issues like stress and depression or the analysis of general mental status. Some studies reported some limitations mainly related to technological issues and lack of standards.

Paper Nr: 81
Title:

A Computational Platform for Heart Failure Cases Research

Authors:

João R. Almeida, Pedro Freire, Olga Fajarda and José L. Oliveira

Abstract: Heart failure is a global health issue that affects millions of people worldwide, and is the main cause of disability and hospitalisation of elderly people. Approximately half of these have heart failure with preserved ejection fraction (HFpEF) and this proportion is increasing as the population ages. There is still no efficient treatment for HFpEF and today’s existing therapies only aim at relieving symptoms. With the aim to unravelling the pathophysiology of HFpEF and identify new therapeutic targets, ongoing long-time studies are collecting patient’s data, including the genomic information. This procedure is complex and requires electronically-stored health information to keep the patient’s information centralised to simplify the following up. In this paper, we present an computational system to support researchers in the different stages of a clinical study, and we describe its use in the management and analyse of HFpEF cohorts.

Paper Nr: 82
Title:

The Need to Optimize the Electronic Health Record: Usability Issues in Legacy Systems Can Compromise Patient Safety

Authors:

Rebecca A. Meehan

Abstract: It is imperative that usability issues affecting patient safety continue to be fixed in the electronic health record (EHR), especially in legacy systems that may not be updated or replaced for years to come. EHR developers and vendors are partners with hospital systems and clinicians in identifying, prioritizing and fixing problems in the EHR that may adversely affect patient safety. Many of these issues are identified by clinicians as issues of poor usability. This presentation discusses current processes for identifying, prioritizing and fixing usability issues as they arise in the implemented or legacy system by both vendors and hospital groups. Strategies for how to improve processes moving forward are discussed.

Paper Nr: 84
Title:

A Mobile Application for Physical Activity Recognition using Acceleration Data from Wearable Sensors for Cardiac Rehabilitation

Authors:

M. Chaari, M. Abid, Y. Ouakrim, M. Lahami and N. Mezghani

Abstract: mHealth applications are an ever-expanding frontier in today’s use of technology. They allow a user to record health data and contact their doctor from the convenience of a smartphone. This paper presents a first version release of a mobile application that aims to assess compliance of cardiovascular diseased patients with home-based cardiac rehabilitation, by monitoring physical activities using wearable sensors. The application generates reports for both the patient and the doctor through an interactive dashboard, as initial proposal, that provides feedback of physical activities of daily living undertaken by the patient. The application integrates a human activity recognition system, which learns a support vector machine algorithm to identify 10 different daily activities, such as walking, going upstairs, sitting and lying, from accelerometer data using a connected textile including movement sensors. Our early deployment and execution results are promising since they are showing good accuracy for recognizing all the ten daily living activities.

Paper Nr: 89
Title:

A Two-stage Imbalanced Learning Method for Sleep Stages Classification using Consumer Activity Trackers

Authors:

Zilu Liang and Mario A. Chapa-Martell

Abstract: Consumer sleep tracking technologies such as Fitbit activity trackers are increasingly used in scientific studies to measure sleep, but these devices are known to be inaccurate for measuring sleep stages. In this study we propose a two-stage imbalanced learning method to improving Fitbit accuracy. The stage-1 model classifies a Fitbit measurement into either correct or incorrect. If the measurement is classified as incorrect, then the stage-2 model corrects it by re-classifying it into one of the four sleep stages. We reliably examined the performance of different combinations of machine learning techniques (i.e. Naive Bayes, random forest and support vector machine) and resampling techniques (i.e. up sampling and down sampling) through leave-one-out nested cross validation. The results showed that using Naive Bayes as the machine learning technique in both stages achieved the best performance, and down sampling needed to be applied during the training of stage-1 model. Our proposed model successfully improved Cohen’s Kappa by up to 27% and Matthews correlation coefficient (MCC) by up to 26%. Performance improvement was achieved mainly through improving the accuracy for light sleep (by 29%). The proposed method can be used to post-process data from Fitbit activity trackers to achieve better accuracy in sleep staging.

Paper Nr: 90
Title:

The Myth of 10,000 Steps: A New Approach to Smartphone-based Health Apps for Supporting Physical Activity

Authors:

Tom Ulmer, Edith Maier and Ulrich Reimer

Abstract: This paper introduces an alternative approach to conventional pedometer apps which measure the wide-spread goal of 10,000 steps a day. Instead we focus on the intensity of physical activity, which is in line with recent recommendations of renowned health institutions such as the WHO. These promote a minimum of moderate to vigorous physically active time per week to achieve the desired health benefits. The paper discusses how the guidelines have been implemented. It also outlines how we help maintain user motivation over time (e.g. by integrating and personalising "nudges") and how we intend to solve the challenges posed by different fitness levels and personal lifestyles.

Paper Nr: 94
Title:

Clinical Performance Evaluation of a Machine Learning System for Predicting Hospital-Acquired Clostridium Difficile Infection

Authors:

Erin Teeple, Thomas Hartvigsen, Cansu Sen, Kajal Claypool and Elke Rundensteiner

Abstract: Clostridium difficile infection (CDI) is a common and often serious hospital-acquired infection. The CDI Risk Estimation System (CREST) was developed to apply machine learning methods to predict a patient’s daily hospital-acquired CDI risk using information from the electronic health record (EHR). In recent years, several systems have been developed to predict patient health risks based on electronic medical record information. How to interpret the outputs of such systems and integrate them with healthcare work processes remains a challenge, however. In this paper, we explore the clinical interpretation of CDI Risk Scores assigned by the CREST framework for an L1-regularized Logistic Regression classifier trained using EHR data from the publicly available MIMIC-III Database. Predicted patient CDI risk is used to calculate classifier system output sensitivity, specificity, positive and negative predictive values, and diagnostic odds ratio using EHR data from five days and one day before diagnosis. We identify features which are strongly predictive of evolving infection by comparing coefficient weights for our trained models and consider system performance in the context of potential clinical applications.

Paper Nr: 101
Title:

Prototypical Implementation of a Decision-supporting System for Operative Breast Cancer Therapy

Authors:

Michael Dück and Eberhard Beck

Abstract: Based on the current edition of the German guideline on Screening, Diagnosis, Treatment and Follow-up of breast cancer, we created a patient journey modelled in BPMN (Business Process Model and Notation V2) serving as template for the development of a patient centered decision support system. This approach resulted in two prototypical devices represented by a web-based information platform and a mobile application, intended to support the decision support at the point of care. These early prototypes were discussed with a clinical expert and the members of a regional breast cancer self-help group. The information gained by this approach will be integrated in the further user centered design of the devices.

Paper Nr: 102
Title:

Analysis of Gaze Trajectory and Skin Extension Pressure Data in Blood Collection Technology

Authors:

Kazuma Mihara, Takeshi Matsuda, Yukie Majima, Seiko Masuda, Masanori Akiyoshi, Kenji Adachi and Naoki Taira

Abstract: Up to this time, research on tacit knowledge of blood collection technology has been conducted, but A method for quantitatively evaluating skills related to blood collection technology and a system that implements them have not been developed. For the present study, using a sensor that can measure eye gaze movement and pressure. Collect finger pressure for skin extension and eye gaze trajectory data during blood collection, and analyze characteristics of pressure distribution during puncture and movement range of the eye gaze and then to examine a method to quantitatively evaluate a part of blood collection technique procedure.

Paper Nr: 103
Title:

The Effects of a Nursing Care Plan Incorporated with a Decision Support System on Ventilator Associated Pneumonia: A Case Study

Authors:

Ozgur Bolat, Nalan Gulenc, Elife Ozkan, Nuran Aydin and Ilker Kose

Abstract: The risk of pneumonia is high in patients who are ventilated in intensive care units (ICUs). Without proper and adequate care, this risk and the mortality rate increases. In a study conducted by the infection committee of our hospital (İzmir Tire State Hospital, the first digital (Stage 7) hospital in Turkey in 2016), it was found that the rate of ventilator-associated pneumonia (VAP) cases increased had increased over three years (2015-17) and was well above the national average. In this study, VAP prevalence in our ICU and the associated extra medication costs were calculated. Furthermore, nursing care plans related to VAP were reviewed and improvements were made according to international standards. The care plan was triggered by criteria defined in a clinical decision support system (CDSS) on the hospital information management system (HIS), and monitorization was conducted to ensure that nurses implement the care plan in a comprehensive and timely manner. As a result of the change, the rate of VAP cases, which had risen to 4.5% in 2017, was reduced to 0.5% in 2018. Similarly, we achieved cost reductions of 90.87% for VAP-based extra medications. Based on these results, it can be suggested that CDSS-supported nursing care can significantly reduce the risk of VAP and increase patient safety in the ICU.

Paper Nr: 110
Title:

Classification of Hand Movement in EEG using ERD/ERS and Multilayer Perceptron

Authors:

Pavel Mochura and Pavel Mautner

Abstract: Continuous EEG activity in the measured subjects includes different patterns depending on what activity the subject performed. ERD and ERS are examples of such patterns related to movement, for example of a hand, finger or foot. This article deals with the detection of motion based on the ERD/ERS patterns. By linking ERD/ERS, feature vectors which are later classified by neural network are created. The resulting neural network consists of one input and one output layer and two hidden layers. The first hidden layer contains 3,000 neurons and the second one 1,500 neurons. A training set of feature vectors is used for the training of this neural network and the back-propagation algorithm is used for the subsequent adjustment of the weights. With this setting and training, the neural network is able to classify motion in an EEG record with an average accuracy of 79.92%.

Paper Nr: 111
Title:

Comparison of Models for Predicting the Risk of Falling in the Non-hospitalized Elderly and Evaluation of Their Performances on an Italian Population

Authors:

Elisa Salvi, Irma Sterpi, Antonio Caronni, Peppino Tropea, Michela Picardi, Massimo Corbo, Giordano Lanzola, Silvana Quaglini and Lucia Sacchi

Abstract: Within the NONCADO project, which aims at preventing falls in the elderly living alone at home, we performed a literature search for models that provide an estimate of the subject’s risk of falling. Our goal is to combine the scores produced by multiple models to derive an overall risk score. In this work we described nine predictive models and we tested their concordance in assessing the risk of falling of two patient populations, namely a simulated patient population and an Italian real-world patient population. Using the real-world population, we also measured the performance of a subset of these models, by comparing their predictions with the outcome (in terms of occurred falls) collected in a 9-months follow-up study. Our experiments showed poor model concordance and dependence of the results on the population. Furthermore, the predictive performance measured the Italian population were limited. Therefore, attempts to combine the risk predictions of multiple models should be cautious.

Paper Nr: 112
Title:

Self-service Data Science for Healthcare Professionals: A Data Preparation Approach

Authors:

Marco Spruit, Thomas Dedding and Daniel Vijlbrief

Abstract: Knowledge Discovery (KD) and Data Mining are two well-known and still growing fields that, with the advancements of data collection and storage technologies, emerged and expanded with great strength by the many possibilities and benefits that exploring and analyzing data can bring. However, it is a task that requires great domain expertise to really achieve its full potential. Furthermore, it is an activity which is done mainly by data experts who know little about specific domains, like the healthcare sector, for example. Thus, in this research, we propose means for allowing domain experts from the medical domain (e.g. doctors and nurses) to also be actively part of the Knowledge Discovery process, focusing in the Data Preparation phase, and use the specific domain knowledge that they have in order to start unveiling useful information from the data. Hence, a guideline based on the CRISP-DM framework, in the format of methods fragments is proposed to guide these professionals through the KD process.

Paper Nr: 113
Title:

Towards Adoption of Standards for Communication Infrastructure/Technologies in Healthcare Systems in LMICs: Theories, Practice and Evaluation

Authors:

Andrew A. Egwar, Richard Ssekibuule and Josephine Nabukenya

Abstract: While electronic health offers great promise to improve healthcare in low and middle-income countries (LMICs), the communication infrastructure/technologies (CI/T) requires standards to improve the current state of none to limited interoperability. This study reviewed theories that inform the assessment of the health system’s readiness to adopt ehealth CI/T standards. The study involved a scoping review of published articles reporting adoption to the use of ICT, technologies, and standards in health. Articles published in English between 2012-2019 were identified through PubMed Central and Google Scholar. Also, grey literature from websites of WHO, standards development organisations and Uganda’s Ministry of Health were searched. Data extraction involved coding to identify key themes that inform the readiness of health systems to adopt standards for eHealth CI. Of the 3,817 published articles, only 32 met the inclusion criteria. 17 grey literature was also included in our discussion. Results identified determinants for eHealth CI/T and that concepts from the technology adoption theories can be used as metrics to assess readiness to adopt standards for ehealth CI/T. The metrics for drivers to adopt standards were higher than inhibitors in Uganda’s health system. The metrics will lead to the development of a readiness assessment framework.

Paper Nr: 115
Title:

Impact of Music on Human Brain Activity during Mental Stress

Authors:

Roman Mouček and Klára Beránková

Abstract: Because music is an integral part of our lives, every step towards a better understanding of its impact on humans is beneficial. This paper deals with the impact of several types of music on the human brain activity during mental stress. An experiment was designed, and electroencephalography data, heart rate data and data from questionnaires were collected, processed and analyzed. All these steps are described and a subset of the large collection of results is presented.

Paper Nr: 118
Title:

“The Algorithm Will See You Now”: Exploring the Implications of Algorithmic Decision-making in Connected Health

Authors:

Noel Carroll, Ita Richardson and Raja M. Abbas

Abstract: Despite abundant literature theorizing on Connected Health innovations to support decision-making, the extant literature provides sparse coverage on users’ awareness of algorithmic decision-making. As a result, little is known regarding the role of algorithmically generated insights which directly influence clinical decisions nor the consequences of distancing clinicians and patients from decision-making capabilities. Indeed, recent studies highlight the growing emphasis on algorithmic decision-making but there is a need to raise questions as to how this is impacting on the risk and quality of delivering care. In this article, a summary of key concerns from the literature is provided, and a discussion on the implications of algorithmic decision-making in Connected Health is presented. In addition, a research roadmap is presented to draw more research focus on the role of algorithmically generated insights in Connected Health.

Paper Nr: 123
Title:

A Hybrid Approach to Develop and Integrate Chatbot in Health Informatics Systems

Authors:

Abhijat Chaturvedi, Siddharth Srivastava, Astha Rai and A. S. Cheema

Abstract: In this paper, we develop a chatbot that seeks free-form natural language queries by its users for blood and related services such as list of blood banks, live blood stock, blood donation camps etc. with one or more parameters as search criteria. The queries can be both Frequently Asked Questions (FAQs) and data driven including location based services. The uniqueness of this chatbot lies in the fact that its architecture provides it flexibility to evolve to encompass more domains and services without having any impact on existing services. Moreover, with approximate keyword initialization, the proposed chatbot can smartly infer from incomplete or incorrect queries by the users as well as has the ability to learn abbreviations. The bot achieves this by leveraging state-of-the-art deep learning and natural language understanding algorithms at the back-end. Specifically, this bot uses a hierarchical approach for parsing queries. At first level, it parses the query into intents i.e. FAQ or data driven. If the classified intent is a FAQ, chatbot to respond while, if it is amongst many of the citizen centric queries, it drills down through the query to identify the entities such as city etc. along with the type of the service and returns the users with the required details.