AI Based Medical Decision Support System Development

OData support
Dr. Strausz György
Department of Measurement and Information Systems

The IT support of healthcare systems is continuously developing, but even today mostly static, database based applications with data recording and search facilities are used. Recent breakthroughs in the field of artificial intelligence are opening up new opportunities in this sector, as well. Beside image processing, analyser and categorizer solutions that are currently used in the diagnostic processes, intelligent decision support systems also became the subject of the developments. These applications can greatly improve the daily work of practitioners and other participants of health services.

In the framework of a physiotherapy research and development project I created a knowledge based medical decision support system to improve the quality and efficiency of the medical diagnostic process. Two expert system solutions were developed in the project. The first one supports patients with musculoskeletal disorders by advising them through the diagnostic process, finding appropriate medical service and by tracking their rehabilitation process, as well. The second solution helps people with body image problems to know more precisely about their condition and to decide on the best cosmetic treatment.

In my work I designed and developed knowledge based models. The model contains all the relevant information form the two subject areas in a computer readable form. I created an application that can use the knowledge model for running an interactive dialog process. The dialog uses predefined questions of the models to gather all the necessary information from the user, and using these data it finds the correct diagnosis. In order to support the whole medical process defined in the project I implemented supplementary functions for the dialog service. As the last step of the development work I designed and implemented user interface to connect the users with the knowledge based.


Please sign in to download the files of this thesis.