A Wavelet Transform Algorithm based Detection and Classification of Sleep Apnea for Monitoring of Health

One-third of a person's life is spent sleeping, and this physiological phenomenon serves several purposes, such as restoring and maintaining normal brain metabolism, allowing the cardiovascular system to recharge, and restoring metabolic balance to the body's glucose supply. Sleep has intimate ties to physical restoration, hormone regulation, and immune system maintenance. During sleep, the body's metabolic rates shift from catabolism (the breakdown of tissues) to anabolism (the rebuilding of tissues). This state of mind is not a constant slumber but rather a dynamic internal structure with a periodic temporal progression. It was often believed that sleep was a non-active state, but modern research has shown that our brains are actually quite busy while we sleep. A multivariate Wavelet transform-based feature selection method has been used since different authors have advocated differing usefulness of different features from the time domain, the frequency domain, and nonlinear features in earlier work on sleep (SL) staging. This provided an optimal collection of 21 features that led to significantly better sleep stage classification performance than previous approaches. Once the signals in the time domain have been transformed into the time-frequency domain, the wavelet coherence coefficients can be calculated to determine the instantaneous respiration rate, and the instantaneous power of the individual signals, the common power, and the phase difference can be examined at the corresponding frequency component. The suggested method improves upon earlier approaches in terms of both event detection and categorization, as evidenced by the experiments.