Automatic Sleep Stages Classification

In this thesis, we first develop an efficient automated classification algorithm for sleep stages identification. Polysomnography recordings (PSGs) from twenty subjects were used in this study and features were extracted from the time{frequency representation of the electroencephalography (EEG) signal. The classification of the extracted features was done using random forest classifier. The performance of the new approach is tested by evaluating the accuracy of each sleep stages and total accuracy. The results shows improvement in all five sleep stages compared to previous works.

Then, we present a simulation decision algorithm for switching between sleep interventions. This method improves the percentage of average amount of sleep in each stage. The results shows that sleep efficiency can be maximized by switching between intervention chains.