Sleep period estimation based on weighted spline smoothing of 24 hour RR interval series

Abstract

It has become feasible to monitor individual heart rate variability for the health care on a daily bases based on the rapid development of IoT sensor networking system. For the precise characterization of the heart rate variability over 24 hours, the segmentation of the data to differentiate sleep period is important. This paper proposes a method of automatic detection of sleep period from 24 hour RRI intervals. Weighted spline smoothing technique has been introduced for the segmentation. Weight function in the spline smoothing optimizing function controls the balance between the data fitness and smoothness. The method adaptively decrease the smoothness factor where the rate of data discrepancy increases. The method enables the natural trend estimation for the accurate segmentation. Computer simulation assuming the presence of additive 1/f fluctuations revealed the effectiveness of the method. The method will be useful for the automatic analysis of Holter ECG big data