Prediction and classification of Human respiratory functions using Flow volume spirometry and radial Basis function neural networks

Abstract

In this work analysis on respiratory mechanics using flow volume newlinespirometry and radial basis function neural network is reported The newlinepulmonary function test using spirometer was recorded from subjects under newlinestudy as per recommended recording protocol The prediction of newlineForced Expiratory Volume in one second FEV was carried out using newlineBackpropagation neural networks Radial Basis Function RBF networks and newlineSelf Organizing Map SOM The SOM generates a set of prototype vectors newlinewhich represent input vector space values These prototypes were used to newlinecreate radial basis function centers The performance of the neural network newlinemodel was evaluated by computing their prediction error statistics of average newlinevalue, standard deviation root mean square and their correlation with the true newlinedata for normal restrictive and obstructive cases Results show that the adopted newline neural networks are capable of predicting FEV1 in both normal and abnormal cases newlinePrediction accuracy was more in obstructive abnormality when compared to restrictive cases newlineThe spirometric data along with the predicted values of FEV1 were used for newlineclassification of normal restrictive and obstructive abnormalities using newlineCombined Neural Networks CNN newline newline