Discovery of rules about complications: A rough set approach in medical knowledge discovery

Abstract

One of the most difficult problems in modeling medical reasoning is to model a procedure for diagnosis about complications. In medical contexts, a patient sometimes suffers from several diseases and has complicated symptoms, which makes a differential diagnosis very difficult. For example, in the domain of headache, a patient suffering from migraine, (a vascular disease), may also suffer from muscle contraction headache(a muscular disease). In this case, symptoms specific to vascular diseases will be observed with those specific to muscular ones. Since one of the essential processes in diagnosis of headache is discrimination between vascular and muscular diseases1, simple rules will not work to rule out one of the two groups. However, medical experts do not have this problem and conclude both diseases. In this paper, three models for reasoning about complications are introduced and modeled by using characterization and rough set model. This clear representation suggests that this model should be used by medical experts implicitly.