( Department of Info. Tech. on Human & Environ. Sci., Grad. Sch. of Sci. & Tech., Kumamoto Univ. )
Keywords: Hypnogram,Quality of sleep,Artificial Neural Network (ANN),Oguri-Shirakawa-Azumi sleep inventory (OSA-SI)
The connection between the psychological evaluation of "quality of sleep" and the physiological one was considered by matching Oguri-Shirakawa-Azumi sleep inventory (OSA-SI) with hypnogram. Eight healthy men (aged 22-23 years old) participated as subjects. They were instructed to go to bed at 0 o'clock (midnight), and to wake up at their own rhythm. As recordings, polysomnography (PSG) was measured during sleeping and OSA-SI was assessed after getting up. This task was performed for three nights by each subject. As analyses, hypnogram was produced from PSG for each night/subject, and then 39 sleep parameter values were calculated from each hypnogram. On the other hand, 5 standardized factor score (convert mean score into 50 for each factor score) were computed from OSA-SI, which was consisted of 16 question items. Finally, artificial neural network (ANN), which was designed to input sleep parameter values and to output factor score (0 means less than 50 points; 1 means more than or equal to 50 points) was employed to estimate "quality of sleep. As results, the best estimation accuracy (71%) was obtained for factor #1 and #2 of OSA-SI. Therefore, it is suggested that factor score of OSA-SI, i.e. "quality of sleep," can be estimated from hypnogram.