Classification of Respiratory Abnormalities Using Adaptive Neuro Fuzzy Inference System

Spirometric evaluation of pulmonary function plays a critical role in the diagnosis, differentiation and management of respiratory disorders. In spirometry, there is a requirement that a large database is to be analyzed by the physician for effective investigation. Hence, there is a need for automated evaluation of spirometric parameters to diagnose respiratory abnormalities in order to ease the work of the physician. In this work, a neuro fuzzy based Adaptive Neuro Fuzzy Inference System (ANFIS), Multiple ANFIS and Complex valued ANFIS models are employed in classifying the spirometric data. Four different membership functions which include triangular, trapezoidal, Gaussian and Gbell are employed in these classification models. Results show that all the models are capable of classifying respiratory abnormalities. Also, it is observed that CANFIS model with Gaussian membership function performs better than other models and achieved higher accuracy. This study seems to be clinically relevant as this could be useful for mass screening of respiratory diseases.