Hybrid system approach to medical diagnosis

Use of the hybrid system approach to medical diagnosis allows all available methodology to contribute to the problem solution. In addition, the modularity of this approach facilitates the development of new applications by replacement of the knowledge base, either in terms of expert-supplied information or from learned knowledge derived from databases. Since the algorithmic components of the system do not change, it is not necessary to re-evaluate these components. The evaluation can therefore focus on the ability of the hybrid system to arrive at accurate decisions for the new application. Standard evaluation methodologies based on sensitivity, specificity, and accuracy must be repeated for each new medical application. The methodology is illustrated in two applications: diagnosis of cardiac disorders and evaluation of dementia.