| Sleep related illnesses are major cause of health and social problems. The cost for the society in terms of unproductive work hours, accidents and health care are enormous. The Medical community is beginning to pay more attention to sleep related disorders, such as sleep apnea, as well as the role and quality of sleep in other conditions.;Obstructive Sleep Apnea (OSA) is a sleep-related breathing disorder affecting about 4% of the population, caused by complete or partial obstruction of airway during sleep. In this dissertation, a method for recognition of the OSA episodes from patients' single channel sleep ECG recording is presented. Automated detection of apnea episodes is facilitated by a machine learning algorithm based on support vector machines classifier.;We have developed two support vector classifiers, one based on subject independent data set and the other based on subject dependent data set. The input data for the support vector classifiers are time and spectral domain features extracted from ECG and ECG derived respiratory (EDR) data. Our contribution also includes the reduction of input data set to the support vector machine using statistical and receiver operator characteristic methods. This reduced data resulted in support vector machine with reduced computational complexity paving way for easy implementation of sleep apnea detector in smartphones and hand-held platforms. The support vector machine algorithms developed here achieved a classification F-measure of 91%.;The second part of the dissertation research is on the development of methods for identifying various sleep states using support vector machines. Our method provides an alternative to the expensive R&K visual scoring method which is commonly used today to assess sleep quality. Our automated sleep quality assessment system measures three sleep quality indices; Sleep Efficiency Index (SEI), Delta-Sleep Efficiency Index (DSEI) and Sleep Onset Latency (SOL). It uses a 3-stage SVM classifier to detect the 4-sleep states: wake (non-sleep), rapid-eye-motion (REM), shallow sleep (stages 1 and 2) and delta deep sleep (stages 3 and 4). |
