A system for automated general medical diagnosis using Bayesian networks.

In this paper we present a computer-assisted diagnostic system for general medical diagnosis developed using Bayesian network methodology and a medical data base created by experts. The system is intended for the general public as a self-diagnostic tool and is available online free of charge (currently only in Polish, with an English version to be released soon). It serves as an educational self-diagnostic tool intended to encourage the user to visit a doctor if the system so suggests, as is most often the case. In this paper we discuss the underlying modeling principles: assumptions behind Bayesian network architecture, solutions to scalability challenges, and computation performance. The distributed software architecture is presented, and finally, initial results based on over 97,000 diagnoses are discussed. The results suggest that the most common health problems for the young generation in Poland (typical user profile) are those resulting from stress and an unhealthy lifestyle.