An Intelligent Diagnosis Flu System Based on Adaptive Neuro-Fuzzy Classifer

Abstract

This study adopts existing three adaptive-neuro-fuzzy classifiers which are neuro-fuzzy classifier with a scaled conjugate gradient algorithm (NFCSCG), neuro-fuzzy classifier with linguistic hedges (NFCLH) and linguistic hedges neuro-fuzzy classifier with selected features (LHNFCSF) to develop an intelligent diagnosis flu system. Gaussian membership function is used for fuzzy set descriptions. Leave-one-subject-out (LOSO) cross-validation is used to estimate the performance of three neuro-fuzzy classifiers. The results shows NFCSCG, NFCLF and LHNFCSF achieved the high accuracy of 100% in the training data. In the testing data, the overall accuracies of LHNFCSF achieved 100%, which is superior to other methods. Thus, this study suggests that LHNFCSF in the intelligent diagnosis flu system can provide a preliminary result to physicians so that the doctor could quickly and accurately decide whether patient have cold or flu.