Abstract:

Sleep is an important time of activity for the daily renewal of our physical and mental health, and it occupies a third of our lives. Sleep disorders can exacerbate or cause symptoms of psychiatric disorders. The first of these may be sleep apnea. Another cause is restless legs syndrome. Depression, anxiety, pain, and some physical problems can also cause insomnia. Sleep apnea can be caused by a nervous system problem or airway obstruction. Studying sleep stages is crucial in diagnosing sleep-related disorders. Sleep stages are also determined by a professional by being with the person during sleep. Considering the average sleep stage diagnosis time of 8 hours, this is quite a long time for a professional. In addition, the definition of sleep stages requires serious expertise and knowledge. The computerized diagnosis system, which automatically diagnoses and treats the diseases described in the literature, has started to be implemented based on theoretical research. This study aims to use deep learning and machine learning techniques to automatically generate sleep stages, which are important parameters in the diagnosis of sleep disorders that directly affect human health. In this study, the random forest algorithm performed the most successful classification (accuracy = 0.974, sensitivity = 0.932, specificity = 0.983). This advanced classification success demonstrates the feasibility of creating a computer-aided diagnostic system that can automatically identify sleep stages, which is an important factor in the diagnosis/treatment of sleep-related disorders.

Keywords: