An investigation of the roles of problem-solving methods in diagnosis

This work centers on dynamic integration of multiple problem-solvers for the purpose of solving problems with broad and complex domains. A problem-solver based on dynamic integration uses diverse problem solving methods (neural net, rule based, model based etc.) to reason about complex problems. This research is based on a preliminary model called TIPS (Task Integrated Problem Solving). TIPS provides a methodology in which the goal structure of a large-grained problem such as diagnosis is mapped to multiple problem-solving methods. A TIPS diagnosis problem-solver has been constructed in the domain of medical diagnosis (liver and blood disorders) which utilizes a number of different problem-solving methods. This paper will discuss the concept of a task-structure, the TIPS architecture and a medical diagnosis system implemented in TIPS.