Performance Analysis of Classifiers in Detection of Respiratory Disorder through Capnography

In this paper the respiratory disorder is detected using classifiers and performance of the classifiers are analyzed. The five statistical features, mean, variance, Skewness, kurtosis and Correlation of PPG signals of PCO2 measurement are considered in this work. The PPG signal features indirectly indicate the condition of heart and lungs. The statistical features of PPG signals are sampled at 200 samples/second for a twenty five epochs are used in this work. The Non Linear Regression (NLR), Gaussian Mixture Model (GMM), Expectation Maximization (EM), and GMM with EM classifiers are used to classify the normal and Respiratory disorder from the PPG signal features. The GMM attained better accuracy of 97.6 % for respiratory case and 94.5 % for normal case compared to other classifiers.