Using Machine Learning Techniques to Improve the Behaviour of a Medical Decision Support System for Prostate Diseases

Abstract

Prostate gland diseases, including cancer, are estimated to be of the leading causes of male deaths worldwide and their management are based on clinical practice guidelines regarding diagnosis and continuing care. HIROFILOS-II is a prototype hybrid intelligent system for diagnosis and treatment of all prostate diseases based on symptoms and test results from patient health records. It is in contrast to existing efforts that deal with only prostate cancer. The main part of HIROFILOS-II is constructed by extracting rules from patient records via machine learning techniques and then manually transforming them into fuzzy rules. The system comprises crisp as well as fuzzy rules organized in modules. Experimental results show more than satisfactory performance of the system. The machine learning component of the system, which operates off-line, can be periodically used for rule updating, given that enough new patient records have been added to the database.