Machine learning helps physicians in diagnosing of mitral valve prolapse

In this paper we present a multimethod approach for induction of a specific class of classifiers, which can assist physicians in medical diagnosing in the case of mitral valve prolapse. Mitral valve prolapse is one of the most controversial prevalent cardiac condition and may affect up to ten percent of the population and in the worst case results in sudden death. MultiVeDec is a general framework enabling researchers to generate various intelligent tools based on machine learning. In this paper we focused on various decision tree methods, which are capable of extracting knowledge in a form closer to human perception, a feature that is very important in medical field. The experiment included classifiers with various classical single method approaches, evolutionary approaches, hybrid approaches and also our newest multimethod approach. The main concern of the latest approach is to find a way to enable dynamic combination of methodologies to the somehow quasi unified knowledge representation. The proposed multimethod approach was capable to outperform all other tested approaches by producing classifier for diagnosing mitral valve prolapse with the highest overall and average class accuracy. More importantly, it was also capable to find some new knowledge important in diagnosing of mitral valve prolapse.