Ebook: Cross-Border Challenges in Informatics with a Focus on Disease Surveillance and Utilising Big Data

This volume contains the proceedings of the 11-th European Federation for Medical Informatics (EFMI, www.efmi.org) Special Topic Conference (STC) held in Budapest, Hungary 27th–29th April 2014. The EFMI STC 2014 is an important international forum for presenting results of current scientific work in health-informatics processes, systems, and technologies.

The EFMI STC is a major scientific meeting. It brings together computer and information science, medicine, and technology. The conference provides a comprehensive overview and in-depth, first hand information on new developments, advanced systems, technologies and applications.

The EFMI STC 2014 was organised by EFMI in collaboration with the Biomedical Section of the John von Neumann Computer Society (NJSZT, further details at: http://njszt.hu/en/neumann/about-njszt). It follows previous conferences in Bucharest, Romania (2001), Nicosia, Cyprus (2002), Rome, Italy (2003), Munich, Germany (2004), Athens, Greece (2005), Timişoara, Romania (2006), Brijuni Island, Croatia (2007), London, UK (2008), Antalya, Turkey (2009), Reykjavik, Iceland (2010), Laško, Slovenia (2011), Moscow, Russia (2012), and Prague, Czech Republic (2013).

The theme of the STC 2014 is “Cross-border challenges in informatics with a focus on disease surveillance and utilising big-data”, addressing a range of important aspects of pan-European and cross-border issues. The conference contains sections providing state of the art presentations on the following theme areas:

• Health data sharing, integration opportunities and challenges across European borders,

• Sources of data, including big data, for monitoring health and disease,

• Using routine data for epidemiological study and public health.

The EFMI Working Groups supporting the conference include “EFMI WG PCI Primary Care Informatics” and “EFMI WG HIIC Health Informatics for Interregional Cooperation”.

The objective of the conference, reflected in the contents of the present volume, was to highlight health-related use of data and the use of these data for monitoring and preventing ill health and disease. They include reports about collaboration at regional, national, and international level. Achieving and maintaining cross-border interoperability of electronic health record systems implies managing the continuous process of change and adaptation of a multitude of elements within and across electronic infrastructures in neighboring countries.

The volume contains twenty-three full papers and poster extended abstracts from specialists of the field. The papers have been selected by the Scientific Program Committee (SPC) out of forty-five submissions of papers, posters, and workshop proposals sent by experts from all over Europe, Russia, Israel and South Africa. Each submission has been reviewed by three reviewers selected from a list of internationally acknowledged domain experts from all continents. The SPC members wish to express their gratitude to the reviewers for the work, which was carried out with the utmost care. All the reviewers' names are listed in the following chapter of these Proceedings.

The topics presented at EFMI STC 2014 are interdisciplinary in nature and consequently of interest to a variety of professionals: medical informatics, bioinformatics, and health informatics scientists, medical computing and technology specialists, public health, health insurance and health institutional administrators, physicians, nurses, and other allied health personnel, as well as representatives of industry and consultancy in the various health fields.

The editors wish to thank the conference patron Dr. Miklós Szócska, Minister of State for Health of the Republic of Hungary. We wish to extended our thanks to Comp-Rend Ltd. for managing the conference and the website (www.stc2014.org), to Thomas Schabetsberger for facilitating the online congress management, and to Bernd Blobel for his caring advice with these Proceedings giving the editors a big part of his vast experience in the field. Finally, we wish to thank the local organizing committee: Dr. István Kósa, István Nagy, Gergely Héja, Gabriella Merth, and Szilvia Tóth-Csuzi.

The editors would like to thank again to all of the authors for their excellent work and the reviewers for their expertise and careful review of the papers.

Lăcrămioara Stoicu-Tivadar, Simon de Lusignan, Andrej Orel,

Rolf Engelbrecht and György Surján

Editors

Extracting scientifically accurate terminology from an EU public health regulation is part of the knowledge engineering work at the European Centre for Disease Prevention and Control (ECDC). ECDC operates information systems at the crossroads of many areas – posing a challenge for transparency and consistency. Semantic interoperability is based on the Terminology Server (TS). TS value sets (structured vocabularies) describe shared domains as “diseases”, “organisms”, “public health terms”, “geo-entities” “organizations” and “administrative terms” and others. We extracted information from the relevant EC Implementing Decision on case definitions for reporting communicable diseases, listing 53 notifiable infectious diseases, containing clinical, diagnostic, laboratory and epidemiological criteria. We performed a consistency check; a simplification – abstraction; we represented lab criteria in triplets: as ‘y’ procedural result /of ‘x’ organism-substance/on ‘z’ specimen and identified negations. The resulting new case definition value set represents the various formalized criteria, meanwhile the existing disease value set has been extended, new signs and symptoms were added. New organisms enriched the organism value set. Other new categories have been added to the public health value set, as transmission modes; substances; specimens and procedures. We identified problem areas, as (a) some classification error(s); (b) inconsistent granularity of conditions; (c) seemingly nonsense criteria, medical trivialities; (d) possible logical errors, (e) seemingly factual errors that might be phrasing errors. We think our hypothesis regarding room for possible improvements is valid: there are some open issues and a further improved legal text might lead to more precise epidemiologic data collection. It has to be noted that formal representation for automatic classification of cases was out of scope, such a task would require other formalism, as e.g. those used by rule-based decision support systems.

To enable the efficient reuse of standard based medical data we propose to develop a higher-level information model that will complement the archetype model of ISO 13606. This model will make use of the relationships that are specified in UML to connect medical archetypes into a knowledge base within a repository. UML connectors were analysed for their ability to be applied in the implementation of a higher-level model that will establish relationships between archetypes. An information model was developed using XML Schema notation. The model allows linking different archetypes of one repository into a knowledge base. Presently it supports several relationships and will be advanced in future.

Immunisation is an important part of health care and adverse events following immunisation (AEFI) are relatively rare. AEFI can be detected through long term follow up of a cohort or from looking for signals from real world, routine data; from different health systems using a variety of clinical coding systems. Mapping these is a challenging aspect of integrating data across borders. Ontological representations of clinical concepts provide a method to map similar concepts, in this case AEFI across different coding systems. We describe a method using ontologies to be flag definite, probable or possible cases. We use Guillain-Barre syndrome (GBS) as an AEFI to illustrate this method, and the Brighton collaboration's case definition of GBS as the gold standard. Our method can be used to flag definite, probable or possible cases of GBS. Whilst there has been much research into the use of ontologies in immunisation these have focussed on database interrogation; where ours looks to identify varying signal strength.

Hypertension, as a cardiovascular risk factor, is a public health issue for which many clinical practice guidelines (CPGs) are published. In order to build the knowledge base of a clinical decision support system for the management of hypertension, we analysed three contemporary (2013) CPGs written in French on hypertension in a knowledge modelling perspective. We developed a semi-automated method using natural language processing called term extraction. Relevant candidate terms have been mapped to medical concepts to produce the conceptual coverage of each CPG. Only 53% of the 88 identified concepts were shared by the three guidelines. While sharing concepts does not warrantee that CPGs contents are similar, this also emphasizes guideline specificities. Such specificities could enrich the knowledge base of decision support systems by broadening their scope.

Employing the bridge between Clinical Information System (CIS) and Clinical Research Environment (CRE) can provide functionality, which is not easily, implemented by traditional legacy EHR system. In this paper, the experience of such implementation at the University Hospitals of Geneva is described. General overview of the mapping of extracted from CIS data to the i2b2 Clinical Data Warehouse is provided. The defined implementation manages to provide the interoperability for the CRE.

An unsolved challenge in biomedical natural language processing (NLP) is detecting ambiguities in the reports that can help physicians to improve report clarity. Our goal was to develop NLP methods to tackle the challenges of identifying ambiguous descriptions of the laterality of BI-RADS Final Assessment Categories in mammography radiology reports. We developed a text processing system that uses a BI-RADS ontology we built as a knowledge source for automatic annotation of the entities in mammography reports relevant to this problem. We used the GATE NLP toolkit and developed customized processing resources for report segmentation, named entity recognition, and detection of mismatches between BI-RADS Final Assessment Categories and mammogram laterality. Our system detected 55 mismatched cases in 190 reports and the accuracy rate was 81%. We conclude that such NLP techniques can detect ambiguities in mammography reports and may reduce discrepancy and variability in reporting.

Building a clinical decision support system (CDSS) capable to collect process and diagnose data from the patients automatically, based on information, symptoms and investigations is one of the current challenges for researchers and medical science. The purpose of the current study is to design a cloud-based CDSS to improve patient safety, quality of care and organizational efficiency. It presents the design of a cloud-based application system using a medical based approach, which covers different diseases to diagnosis, differentiated on most important pathologies. Using online questionnaires, traditional and new data will be collected from patients. After data input, the application will formulate a presumptive diagnosis and will direct patients to the correspondent department. A questionnaire will dynamically ask questions about the interface, and functionality improvements. Based on the answers, the functionality of the system and the user interface will be improved considering the real needs expressed by the end-users. The cloud-based CDSS, as a useful tool for patients, physicians and healthcare providers involves the computer support in the diagnosis of different pathologies and an accurate automatic differential diagnostic system.

The paper presents a fuzzy inference system based prediction with the role to determine the appropriate action for patients that presents lower back pain. If not treated correctly lower back pain can degenerate in various diseases. The system infers three possible actions: (1) spinal cord surgery, (2) medication combined with exercises and (3) no action needed. The system takes in consideration the age and sex of the patient, a pain intensity parameter, the metabolic rate of the patient and mobility parameters from the Zebris Mobility device. In total 243 rules have been formulated but only 21% of the rules suggests surgery. The initial results are promising; there is a correlation of 0.83% between the control results and the results from the system.

The paper describes the first, preclinical evaluation of a dietary logging application developed at the University of Pannonia, Hungary. The mobile user interface is briefly introduced. The three evaluation phases examined the completeness and contents of the dietary database and the time expenditure of the mobile based diet logging procedure. The results show that although there are substantial individual differences between various dietary databases, the expectable difference with respect to nutrient contents is below 10% on typical institutional menu list. Another important finding is that the time needed to record the meals can be reduced to about 3 minutes daily especially if the user uses set-based search. Conclusion: a well designed user interface on a mobile device is a viable and reliable way for a personalized lifestyle support service.

The Israel National Hospital Discharge Register (INHDR) is an essential section of healthcare data. It includes record for each admission to hospital wards during the last twenty years, and the data are increasing by digitally updated information from hospitals on continually a monthly or quarterly basis. The register contains encrypted patient identity number, admission number, demographic and geographic data, hospitalization data, diagnoses, procedures and accounting data. The goal of the register is to measure medical and surgical services in hospitals, to compare hospital activity among regions, gender and age and population groups within the country and among other countries, to analyse the difference between periods. This large-scale hospital data helps in planning of the hospital services, analysing the health status of the population, disease and injury surveillance, and helps in performance of quality indicators. It assists decision makers at the Ministry of Health (MOH) in their daily and on-going missions.

With information and data in all domains growing every day, it is difficult to manage and extract useful knowledge for specific situations. This paper presents an integrated system architecture to support the activity in the Ob-Gin departments with further developments in using new technology to manage Big Data processing – using Google BigQuery – in the medical domain. The data collected and processed with Google BigQuery results from different sources: two Obstetrics & Gynaecology Departments, the TreatSuggest application – an application for suggesting treatments, and a home foetal surveillance system. Data is uploaded in Google BigQuery from Bega Hospital Timişoara, Romania. The analysed data is useful for the medical staff, researchers and statisticians from public health domain. The current work describes the technological architecture and its processing possibilities that in the future will be proved based on quality criteria to lead to a better decision process in diagnosis and public health.

Successful service integration in policy and practice requires both technology innovation and service process innovation being pursued and implemented at the same time. The SmartCare project (partially EC-funded under CIP ICT PSP Program) aims to achieve this through development, piloting and evaluation of ICT-based services, horizontally integrating health and social care in ten pilot regions, including Kraljevo region in Serbia. The project has identified and adopted two generic highest-level common thematic pathways in joint consolidation phase – integrated support for long-term care and integrated support after hospital discharge. A common set of standard functional specifications for an open ICT platform enabling the delivery of integrated care is being defined, around the challenges of data sharing, coordination and communication in these two formalized pathways. Implementation and system integration on technology and architecture level are to be based on open standards, multivendor interoperability, and leveraging on the current evolving open specification technology foundations developed in relevant projects across the European Research Area.

Treatment of diabetes mellitus is a public health related problem of modern healthcare. Surveys show that current methods to estimate the required amount of insulin are quite inefficient in practice as they are based on experience. This paper offers a new approach to predict the glucose level of people with diabetes. It combines two efficient models of the literature: one for nutrient absorption and one for glucose control. The combination of them tracks the blood sugar level considering nutrition composition, applied insulin and initial glucose level. Compared to already existing mixed meal models, the current version takes into account a more detailed nutrition composition (protein, lipid, monosaccharide, fiber and starch) supported by our expert dietary systems. Although the model gives satisfactory results even with parameter sets taken from literature, parameter training by genetic algorithms yields a better tracking of the patients.

Starting in 2009, the first ever Belgian nationwide data collection network using routine data extracted from primary care EPR (upload method) has been built from scratch. The network also uses a manual web-based data collection method. This paper compares these two methods by analysing missing and most recent values for certain parameters. We collected data from 4954 practices, pertaining to 29,180 patients. Mean values for the most recent parameters were similar regardless of which data collection method was used. Many missing recent values (>46%) were found for all of the parameters when using the upload method. It seems that, in Belgium, uploading routine data from primary care EPR on a large scale is suitable and allows the collection of chronological retrospective data. However, the method still requires major, carefully controlled improvements.

In this paper we propose a new Classification based on Association Rules (CAR) algorithm that improves the interpretability of the results, works over real data from the electronic health records (EHRs), and allows the study of the patient as a whole. It enables tasks such as the discovery of relationships between diseases, or offering several alternative and reasoned diagnoses for the cases of patients with several diseases that analysed separately could lead to mistaken diagnosis. We aim to achieve several goals: to discover hidden relationships; to improve the interpretability and reduce the complexity of the result; to obtain more reliable diagnosis (getting alternative reasoned diagnoses and higher robustness to noisy rules), and to improve the quality of the classifier avoiding the usual over-fitting problem. To this purpose, we define and exploit hierarchies defined over datacubes dimensions, and change the way the association rules are obtained, and their evaluation at the classification process. To prove the utility of our proposal we have used it in an example of cancer discrimination.

The paper describes the investigation of the Hungarian public administrative health databases with the aim to identify hidden correspondences in the patients' evaluation pathways for patients with suspected coronary artery disease (CAD). In our current work we investigated the effect of the waiting times of invasive and non-invasive investigations in the evaluation pathways of patients with suspected CAD. We found a considerable correlation between waiting times and the further course of the patients.

The research team needed to upsize the solution previously tested so that it could expand the routine data collected via tablet computers. The research team identified the general flow of data within clinics. Data was mainly collected from registers, which were later converted to electronic form and checked for duplication. A database was designed for the collection of demographic data (Patient Master Index), which was aimed at eliminating duplication of patients' data in several registers. Open Data Kit (ODK) Collect was setup on Android tablets for collecting disease related routine data, while ODK Aggregate as the storage and aggregates of data captured by ODK Collect and the Patient Master Index for demographic data, were setup on an Apple Mini Mac server. Data collection is in progress. The expected results include improved data quality, reliability and quick access to summary data. Secondly, instant retrieval of patient demographic details and clinic numbers are included. Thirdly, ability to form standard reporting from the SQL database and lastly exporting data into the TIER.net and DHIS systems via CVS files thus eliminating the need for data capturers are shown.

Due to the need for an efficient way of communication between the different stakeholders of healthcare (e.g. doctors, pharmacists, hospitals, patients etc.), the possibility of integrating different healthcare systems occurs. However, during the integration process several problems of heterogeneity might come up, which can turn integration into a difficult task. These problems motivated the development of healthcare information standards. The main goal of the HL7 [1] family of standards is the standardization of communication between clinical systems and the unification of clinical document formats on the structural level. The SNOMED CT [2, 3] standard aims the unification of the healthcare terminology, thus the development of a standard on lexical level. The goal of this article is to introduce the usability of these two standards in Java Persistence API (JPA) [4] environment, and to examine how standard-based system components can be efficiently generated. First, we shortly introduce the structure of the standards, their advantages and disadvantages. Then, we present an architecture design method, which can help to eliminate the possible structural drawbacks of the standards, and makes code generating tools applicable for the automatic production of certain system components.

Paper presents an overview of the EU funded Project of Curriculum Development for Interdisciplinary Postgraduate Specialist Study in Medical Informatics named MEDINFO to be introduced in Croatia. The target group for the program is formed by professionals in any of the areas of medicine, IT professionals working on applications of IT for health and researchers and teachers in medical informatics. In addition to Croatian students, the program will also provide opportunity for enrolling students from a wider region of Southeast Europe. Project partners are two faculties of the University of Zagreb – Faculty of Organization and Informatics from Varaždin and School of Medicine, Andrija Štampar School of Public Health from Zagreb with the Croatian Society for Medical Informatics, Croatian Chamber of Economy, and Ericsson Nikola Tesla Company as associates.