INTELLIGENT TECHNIQUES FOR DATA INTEGRATION AND DECISION SUPPORT IN THE MEDICAL DOMAIN

摘要

Data commonly produced during medical practice often falls into the category of complex data, which may not comply with the traditional field structure of common types like alphanumeric, dates, etc., with prominent examples including text documents, time-series, and multimedia objects. In addition, medical data may not be mutually compatible and easy to integrate into a centralized repository, since it originates from heterogeneous sources which apply different assumptions, conventions and schemas. Having in mind these challenges, two research teams from Humboldt University, Berlin, Germany and University of Novi Sad, Serbia, organized the joint research project "Intelligent Techniques for Data Integration and Decision Support in the Medical Domain," with the aim of investigating and applying different techniques for reasoning, mining and retrieval to problems recognized in real-world scenarios within the medical domain. This paper describes the initial research efforts of the two groups towards exploring how to overcome difficulties in organizing different kinds of medical data, incorporating case-based reasoning/time-series techniques in the implementation of a reliable general-purpose decision-support framework, and generating special-purpose decision-support systems from the general framework. We expect the proposed research to provide scientific results and facilitate the implementation of appropriate intelligent software tools applicable in different domains as standalone applications, but also as components that can be integrated into already existing information systems and environments.