Research

Scientific and research activities of the EuroMISE Centre are focused on solving research problems not only in the field of informatics and statistics but also in biomedical fields. These activities are usually supported by Czech and European grants in cooperation of the Department of Medical Informatics ICS AS CR with other national and foreign institutions.

Involvement in several European projects aimed at working with uncertainty in medicine (MUM), formalization of medical guidelines (MGT) and multimedia open electronic record about a patient with applications to cardiology (I4C-TripleC) helped with the scientific orientation of the EuroMISE Centre in the field of biomedical informatics and statistics and it helped also with the development of chosen fields of research. From many Czech projects we can mention at least thematically important projects connected with the field of cardiology, genetics and epidemiology, which were solved with the support of the Czech Science Foundation during the years 1993-1995 (Design of methods for the data analysis in epidemiological studies, searching for risk factors and modelling of disease risks ) and with the support of the Internal Grant Agency of the Ministry of Health CR during the years 1997-1999 (Long-lasting study of 40-50 years old men undergoing 20 years of treatment - impact on mortality and morbidity) and during the years 1998-2000 (Moderate hyperhomocysteinemy in the Czech population: the analysis of genetic factors in patients with atherosclerosis). Besides applications in cardiology the activities of the EuroMISE Centre are focused on applications in other fields, such as rheumatology, neurology, neurosurgery, gynaecology and obstetrics, urology, pediatrics, oncology, ophthalmology, general medical practice, forensic medicine, and criminological practice. The activities of the EuroMISE Centre on the national level and within interdisciplinary research cooperation are focused mainly on problems, methods and tools for solving of important biomedical and healthcare tasks. These are research in the field of the electronic health record, systems for decision making, methods for data mining from biomedical databases and knowledge bases, advanced statistical methods and methods of information theory for applications in biomedicine.

Looking for new forms of electronic health documentation, especially in connection with issues of integrating electronic records to the system of Czech heath care in relation to the Czech legislation and Czech national environment, was one of the first steps in the research of new technologies and systems for structured health documentation management. Executive and legislative data are collected for the development of national data environment of healthcare documentation system and intelligent telemedicine systems. The key standpoint in research and development is an effort for compatibility and interoperability with European and international environments. New trends of electronic health documentation, which are promoted especially in research projects in Europe and USA, include also working with medical knowledge. It is a research of new ways of knowledge representation in multimedia electronic health records. In this field the EuroMISE Centre continues in the experience gained during the European projects I4C-TripleC and MGT and it works on chosen European standards. The research is developed with the aim to design and describe the system for electronic health documentation management, which would respect European standards and which would be adjusted to the Czech national environment and which would bring also a unified perspective to structurally stored data and control mechanisms over entered data. This would be based also on medical guidelines and a user-friendly and easy to handle environment. During the years 2000-2004 in the framework of the LN00B107 project of the Ministry of Education CR a proposition for a universal dynamic structure for representation and storing of health data was suggested and the architecture of electronic health record software realization was designed and named MUDR (MUltimedia DistRibuted record). Further, the MUDRLite application was proposed, which reacted directly on requirements to develop a simple version of the electronic health record, which would be able to work on relatively little powerful working stations without any connection to a remote database server. MUDRLite was applied in dental medicine and in neurosurgery in cooperation with the Department of Stomatology and the Department of Neurosurgery of the First Faculty of Medicine, Charles University in Prague. For example, the knowledge base for dental medicine was modelled and implemented into the electronic health record. In the year 2005 the patent application on the structured graphic component of the dental cross was deposited.

Medical guidelines are professional medical texts concerning procedures of diagnostics, therapies or prevention of individual diseases or their groups. They can be used for standardization of a particular disease treatment, improvement of patient care by introducing new processes, for quality comparison of patient care in individual medical institutions, for teaching healthcare workers, for development of applications of some medical operations, for design of decision support systems, for improvement of interdisciplinary cooperation in medicine, etc.

Research activities in this field were focused on general issues of medical knowledge formalization and formalization of knowledge included in medical guidelines. By formalization of textual medical guidelines we understand extraction of knowledge included in them into the electronic and machine processible form. The procedural structure of guidelines and the list of significant parameters are gained by the formalization. The graphic GLIF model is the result of formalization (GuideLine Interchange Format - it originated from the cooperation of University Columbia, Harvard, McGill and Stanford universities, see http://www.glif.org). The passage through individual branches of the GLIF model depends on fulfilment or non-fulfilment of requirements in decision steps or on the interactive selection of a relevant graph branch by a user. As the GLIF model is graphic, it is necessary to encode it in XML (eXtensible Markup Language) before further processing. More information about research results can be found in publications and demonstrations of chosen medical guidelines (zde dát odkaz na část tohoto webu lékařská doporučení).

Statistics in biomedical research is a field, which affects all of us in a specific way. Medicine brings many interesting and often difficult problems, which lead in statistics to more or less sophisticated methods and models. With physicians we try to analyze and interpret detected conclusions from clinical experiments and epidemiological studies and to contribute partially to the development of medicine, not only as a science field but also to the improvement of health care itself. The main aim of statistical methods in the medical research is to secure correctness and proficiency of statistical data evaluation and interpretation of acquired results. The EuroMISE Centre covers a wide spectrum of statistical methods applications in the biomedical research. The emphasis is put mostly on the statistical design of medical studies whose theme is to study prevention of civilizational diseases incidence. Cardiovascular diseases, metabolic disorders, allergies manifestations, incidences of tumorous diseases, etc. are examples of civilizational diseases. Statistical methodologies are used in designing of systems for decision making, e.g. in relation to evaluation of risk factors in a monitored patient from the view point of some disease incidence, in relation to evaluation of genetic predispositions from the view point of monitored disorders or anomalies incidence. Here we show some examples of data analyses developed by the EuroMISE Centre. The EuroMISE Centre also provides statistical consultations for physicians, the Centre helps with the design of experiments and epidemiological studies, it contributes to designing and developing of suitable questionnaires for data collections, to selection of appropriate methods for data analyses, results interpretation and preparations of papers for professional journals, presentations in scientific conferences and preparations of research reports.

To fulfil aims in the research field of structured health documentation, it was necessary to create specific areas of health care for which a database model was designed and for which data collection proceeded. In cooperation with computer scientists and statisticians the physicians, especially from the General University Hospital in Prague and Municipal Hospital in Caslav, prepared the procedure for a uniform collection of basic data in cardiology. The result is the Minimal Data Model for Cardiology (MDMC). The single-purpose application ADAMEK was developed on bases of the prepared minimal data model for cardiology. Formation of data models gradually improved and new software tools for facilitating of data model formation, for easy computations of gathered data, e.g. the DATAMO software system, were developed. To find new knowledge from information stored in biomedical databases and knowledge bases, the research was focused on various techniques of extraction and visualization of new knowledge. The process of knowledge mining is described in more details in outputs of the University of Economics in Prague, which is one of the workplaces of the EuroMISE Centre (http://euromise.vse.cz/kdd/index.php?page=proceskdd). One of the applications of knowledge data mining in cardiology is the analysis of an extensive epidemiological study of primary prevention of atherosclerosis (http://euromise.vse.cz/stulong/index.php).

In the field of automatic extraction of structured information from medical reports written in the form of a free text the EuroMISE Centre during the years 2000-2004 with the support of the LN00B107 project examined methods based on the analysis of regular rules. The analysis of medical reports from the Municipal Hospital in Caslav, from the 2^nd and 3^rd Department of Internal Medicine of the First Faculty of Medicine, Charles University in Prague and General University in Prague was carried out. Further, the reports were analyzed morphologically by means of a lemmatizer and by means of morphologically marked texts the frequency lexicons were compiled. Successfulness of the regular analysis was shown in the analysis of anonymized medical records from the Municipal Hospital in Caslav and General University Hospital in Prague.