Heart rate-based nighttime awakening detection

Sleep fragmentation is a cause of impaired daytime performance. We have developed an algorithm for detection of nighttime awakenings based on heart rate. As much as 15 healthy normal sleepers, 23 +/- 3 years, participated in this study. The dataset contains 33 nights of polysomnographic (PSG) and electrocardiographic (ECG) measurements. After a habituation night, the subjects underwent a reference night without interventions, followed by some nights with interventions. These included noise, light, physical and cognitive interventions. Nighttime awakenings were subdivided in to awakenings (>15 s) and short awakenings (<15 s). The overall number of awakenings was 18.5 (+/-10.5) and short awakenings 13.2 (+/-10.5). The number of nighttime awakenings did not differ significantly between the reference and intervention nights; a repeated measures ANOVA resulted in a p value of 0.66 for awakenings and 0.57 for short awakenings. As much as 5 reference nights were used as training set, 28 as validation set. The algorithm detects the awakening periods with a sensitivity of 80.5% (confidence interval 77.9-82.9%). Heart rate is an adequate measure that allows for detection of nighttime awakenings and hence sleep quality.