Critical Slowing Down in Heart Rate Variability for Human Condition Control: An Example of Sleep Onset Detection

Modern technologies offer numerous opportunities for detecting the human condition. The practical implementation of such solutions is relevant both for various industries and for virtual activities—wherever they are associated with a significant cognitive load and a high risk of errors due to human factors, such as loss of vigilance from fatigue or drowsiness. The research presented in this paper validates an approach for the rapid detection of sleep onset by analyzing a single metric of heart rate variability, utilizing the concept of critical slowing down. The material for this research consisted of 4 evening-night recordings of NN intervals, where moments of sleep onset were marked for each participant. Standard deviations (SDNN) and autocorrelation coefficients of NN intervals were analyzed within sliding windows. It was found that immediately after sleep onset, there was a sharp and pronounced decrease in both metrics, regardless of the time of falling asleep, with the dynamics of SDNN proving to be more indicative. The results of this pilot study can be utilized for further exploration of early warning signals in heart rhythm indicators that would indicate a loss of vigilance. This may serve as a foundation for the development of systems that predict a decline in cognitive control due to sleepiness during natural activities.