Effect of the continuity and sampling rate of polygraphy data on sleep stage by a computerized scoring system--a reliability study.

Abstract

The percentage of each sleep stage determined by visual judgment was compared with computerized judgment using continuous and three intermittent sampling data. Comparison of stage agreements of four types of computerized judgments with visual judgment were also conducted. Hypnograms of six healthy young male students were recorded on each of five nights. Data of each subject were compiled individually, and the six data were analyzed. Agreement rate of the sleep stage by computerized judgment using continuous (method A), one-third (method B), one-ninth (method C), and one-fifteenth (method D) data with that of visual judgment were compared. The agreement rate of visual and computerized judgment using method A showed no difference from that between visual and computerized judgment using method B. There was a significant loss of agreement between visual and computerized judgment using method C or D compared with that between visual and computerized judgment using method A. Paired t-tests were conducted for each stage percentage against total sleep time of visual judgment and those of computerized judgments using method A, B, C, or D. There were no significant changes in any sleep stage percentages between visual judgment and computerized judgments using method A, B, C, or D. From sleep stage agreement and the degree of difference from visual judgment, the acceptable limitation of the sampling period was concluded to be one-third of continuous data.