# Exploring the relationship between health behavior and sleep quality: preliminary insights from ECG-derived sleep analysis

## TL;DR

This study explores how health behaviors like physical activity and diet relate to sleep quality using a new ECG-based sleep analysis method.

## Contribution

A novel algorithm for determining sleep stages using ECG data is introduced and validated against health behavior and metabolic variables.

## Key findings

- Moderate physical activity and fruit/vegetable consumption were positively linked to sleep stages.
- BMI showed negative correlations with sleep stages.
- Health behavior variables predicted NREM and REM sleep duration and episodes but not total sleep duration.

## Abstract

Correlations between sleep quality, health behavior, and metabolic variables are empirically well documented. This study investigated a novel algorithm to determine sleep stages by leveraging ECG data. The aim was to examine whether this algorithm effectively capture correlations with health behavior, self-reported sleep quality, and metabolic variables. A cross sectional correlation study design was used to survey health behavior of a total of 194 healthy individuals (87 female; mean age, 40.29 ± 11.8) with a focus on BMI, physical activity, subjective sleep quality via the PSQI and eating habits (daily fruit and vegetable consumption). In addition, a 24 h ECG was derived to determine sleep stages. Positive associations were found between sleep stages, sleep duration and sleep quality. Moderate physical activity and fruit and vegetable consumption were positively associated with sleep stages. Negative correlations were found between BMI and sleep stages. Overall, health behavior variables could predict duration of NREM (R2 = 0.040, F(4/186) = 2.98, p < .021) and REM (R2 = 0.050, F(4/188) = 3.50, p < .009), as well as numbers of NREM and REM episodes (R2 = 0.065, F(4/186) = 4.30, p < .002; R2 = 0.077, F(4/186) = 4.99, p < .001). However, health behavior variables could not predict sleep duration (R2 = 0.010, F(4/168) = 1,44, p < .222). The SleepECG algorithm supported correlations between health behaviors, BMI, and self-reported sleep quality, indicating its potential as a practical and cost-effective method for objectively measuring sleep when polysomnography is not feasible.