Estimation of sleep stage in the falling asleep period using a Lorenz plot of ECG RR intervals.

The falling asleep period is the shift from the waking stage to sleep stages 1 and 2. Changes during the falling asleep period can be observed on electroencephalograms (EEGs). In this research, we developed a technique for estimating sleep stage at the falling asleep period without using EEGs. We performed a Lorenz plot (LP) using the intervals between heartbeats, known as electrocardiogram (ECG) RR intervals, of the falling asleep period, and confirmed that changes in the distribution on the LP occur according to changes in sleep stage. To evaluate the changes in these distributions, we projected the LP on y = x axis and y = -x axis, and analyzed the shifting of mean and standard deviation in each sleep stage. The results demonstrated that the distance from the coordinate origin to the mean of distribution became longer as sleep stage deepened, but the variations in the distribution of the LP stabilized. By quantitatively evaluating these phenomena, we proposed two indices of mean (M of LP) and ellipse area (S of LP) of the falling asleep period. Additionally, a multiple regression analysis was done to calculate sleep stage quantitatively, culminating in the derivation of the estimated equation of falling asleep period. Therefore, we could easily estimate the directionality of the sleep stage at the falling asleep period using a LP of ECG RR intervals.