A hybrid case-based medical diagnosis system

This paper proposes a hybrid case-based system to help the physician. It includes a hypermedia human-machine interface and a hybrid case-based reasoner. The hypermedia human-machine interface provides a friendly human body image map for the clinician to easily enter a given consultation. It utilizes a medicine-related commonsense knowledge base to help complete the input data during the consultation. The hybrid case-based reasoner is responsible for selecting and adapting relevant cases from the case library into a diagnosis for the consultation. This reasoner does those jobs by hybridizing many techniques. Basically it uses a distributed fuzzy neural network for case retrieval. It employs decision theory, constrained induction trees, and relevance theory for case adaptation involving case combination. The technique is also used for learning new cases into the case library. Hybridizing these techniques together can effectively produce a high quality diagnosis for a given medical consultation.