A way to enhance the performance of a model that combines genetic algorithmsand fuzzy logic for feature selection and classification is proposed. Earlydiagnosis of any disease with less cost is preferable. Diabetes is one suchdisease. Diabetes has become the fourth leading cause of death in developedcountries and there is substantial evidence that it is reaching epidemicproportions in many developing and newly industrialized nations. In medicaldiagnosis, patterns consist of observable symptoms along with the results ofdiagnostic tests. These tests have various associated costs and risks. In theautomated design of pattern classification, the proposed system solves thefeature subset selection problem. It is a task of identifying and selecting auseful subset of pattern-representing features from a larger set of features.Using fuzzy rule-based classification system, the proposed system proves toimprove the classification accuracy.
展开▼
