Evaluation of forced expiratory volume prediction in spirometric test using Principal Component Analysis

In this work, an attempt has been made to evaluate the clinical relevance of lung function spirometric test using neural network based prediction and Principal Component Analysis (PCA). Flow-volume curves generated using spirometer on subjects (N = 175) under standard recording protocol were used for the present study. The most significant spirometric parameter FEV1 predicted using Radial basis function neural networks (RBFNN) incorporating k - means clustering method was considered further for analysing the interdependency of spirometric parameters. PCA was performed on the measured and predicted datasets to analyse the interdependency among the parameters. Results demonstrate that PCA confirms the consistency in prediction of FEV1 for normal subjects. It appears that this method of prediction and principal component analysis could be useful in explaining the significance of FEV1 in assessing the lung function abnormalities for spirometric pulmonary function test with incomplete data. © 2011 Inderscience Enterprises Ltd.