Classification and Diagnosis of Sleep Apnea and Congestive Heart Failure Using DWT Based Statistical Features: An Extensive Study

Electrocardiogram assumes a significant part in the diagnosis of various cardiovascular disorders. This work deals with a comprehensive method to extract various statistical features from ECG signal to diagnose two major cardiovascular disorders i.e. Sleep Apnea and Congestive Heart Failure. Sleep Apnea is a serious sleep disorder that occurs when a person's breathing is hindered during sleep. Congestive heart failure (CHF) is a chronic condition that affects the pumping power of a person's heart muscles. In the present work, the principle of Discrete Wavelet Packet decomposition is applied to diagnose Sleep Apnea and Congestive Heart Failure from the background signal. Two different Mother Wavelets I.e. Biorthogonal and Daubechies wavelets are used to extract statistical parameters like Mean, Median, Standard Deviation, Energy and Approximate Entropy(ApEn). These parameters are extracted and compared, from the ECG signal of a healthy person and a person suffering from Sleep Apnea. A similar procedure is repeated for the ECG signal of a person suffering from Congestive Heart Failure. Significant contrasts have been found between ApEn, energy and standard deviation values of individuals experiencing Sleep Apnea and also Congestive Heart Failure. Various machine learning algorithms are used to train the classification models for classifying the disorders as mentioned above. Clear distinguishable values are observed in the trends obtained by both the wavelets, yet Daubechies is seen to perform better. The inclusion of many parameters has helped build up the accuracy with even an outstanding 100% using tree classifier.