Assessment and classification of normal and restrictive respiratory conditions through pulmonary function test and neural network.

In this work, an attempt to classify respiratory abnormality using a pulmonary function test and neural networks is reported. The flow - volume curves generated by spirometric pulmonary function tests were recorded from subjects under study. The pressure and resistance parameters were derived using theoretical approximation of the activation function representing the pressure - volume relationship of the lung. The pressure - time and resistance - expiration volume curves were obtained during maximum expiration. The derived values together with spirometric data were used for classification of normal and restrictive abnormality using feed forward network. Results demonstrate the ability of the proposed method in identifying and classifying pulmonary function data into normal and restrictive cases. The validity of the results was confirmed by measuring accuracy (92%), sensitivity (92.3%), specificity (91.6%) and adjusted accuracy (91.95%). As spirometric evaluation of human respiratory functions are essential components in primary care settings, the study carried out seems to be clinically relevant.