Automatic Sleep Stage Determination by Probabilistic Method Based on Experts' Knowledge

Abstract

Determination of sleep stages for recorded polysomnogram is an important in clinical diagnosis such as sleep disorders. Visual inspection of polysomnogram consumes much time and is a laborious task. Therefore, many researches on automatic determination of sleep stages have been done in the past. In this study, automatic determination of sleep stages by using the multi-valued decision making method based on conditional probability was proposed. The method was constructed according to the database inspected by an expart and the manner of visual inspection. Proposed method was adopted to the EEG data acquisited from 4 patients, and evaluated by comparing the result of visual inspection. Proposed method was usable for determination of sleep stages, and will be an effective tool for clinical diagnosis.