Extraction of Risk Factors by Multi-agent Voting Model Using Automatically Defined Groups

In medical treatment, it is difficult to diagnose diseases directly from raw real value data obtained by medical examination for patients. For support of the task, it is necessary to transform the raw data into meaningful knowledge representations, which represent whether the characteristic symptoms of the disease are observed. In this research, we aim to acquire such multiple risk factors automatically from the medical database. We consider that a multi-agent approach is effective for extracting multiple factors. In order to realize the approach, we propose a new method using an improved Genetic Programming method, Automatically Defined Groups (ADG). By using this method, multiple risk factors are extracted, and the diagnosis is performed through multi-agent cooperative voting. We applied this method to the coronary heart disease database, and showed the effectiveness of this method.