Integration of Causal Models into Medical Expert Systems

The Project

The goal of this project is to identify principles and basic methods for reasoning about causal and qualitative relations. Special emphasis is put on the integration of surface knowledge acquired by experts from years of experience and of deep knowledge giving a partial model of the domain. The current focus of the project is model-based diagnosis for rheumatology. The diagnostic decision-support system MESICAR utilizes detailed anatomical knowledge about the musculoskeletal system. MESICAR-LEARN extends the knowledge of MESICAR by learning disease descriptions of specialized diseases frequently seen with the help of machine learning techniques.

MESICAR is an experimental decision-support system for rheumatology. It incorporates a detailed representation of the anatomy in order to be able to reason about causal relationships between disturbances affecting the musculoskeletal system. This yields two main advantages: 1) it enabled us to build generic disease descriptions. Instantiation automatically constructs specific disease descriptions by filling in the anatomical details which describe the situation of the patient; 2) the system provides a user interface showing all the anatomical details within the context of the patient's problem. This is essential for the intended field of application, namely, primary medical care.

The Project Team

Werner Horn
Kurt Ammer (consulting rheumatologist)
Gerhard Widmer (MESICAR-LEARN)
Bernhard Nagele (MESICAR-LEARN).

Publications

Horn W.: MESICAR - A Medical Expert System Integrating Causal and Associative Reasoning, Applied Artificial Intelligence, Special Issue on Causal Modeling, 3(2-3)305-334, 1989.
Horn W.: Diagnostic Decision Support Based on Generic Disease Descriptions and Detailed Anatomical Knowledge, in Hunter J., et al.(eds.), Proceedings of the Second European Conference on Artificial Intelligence in Medicine (AIME 89), Springer, Berlin, pp.299-308, 1989.
Horn W.: Integration anatomischen Wissens in ein rheumatologisches Expertensystem, in Wetter T., et al.(eds.), GMDS/GI-Workshop "Wissensbasierte Systeme in der Medizin", IBM Heidelberg, Institut für Wissensbasierte Systeme, IWBS-Report 112, 1990.
Horn W.: Utilizing Detailed Anatomical Knowledge for Hypothesis Formation and Hypothesis Testing in Rheumatological Decision Support, Artificial Intelligence in Medicine, 3(1)21-39, 1991.
Horn W., Widmer G., Nagele B.: Learning Specialized Disease Descriptions in a Rheumatological Expert System, in Adlassnig K.-P., et al.(eds.), Medical Informatics Europe 1991 (MIE-91), Springer, Berlin, pp.322-326, 1991.
Horn W., Widmer G., Nagele B.: Learning Specialized Disease Descriptions in a Rheumatological Expert System. In Medical Informatics Europe, MIE-91 (eds. K.-P. Adlassnig et al.), Springer Verlag, Berlin 1991.
Widmer G., Horn W., Nagele B.: Automatic Knowledge Base Refinement: Learning from Examples and Deep Knowledge in Rheumatology. Artificial Intelligence in Medicine, 5(3)225-243, 1993 (see also OFAI-TR-92-16).