Abstract
Consumer wearables compute overnight heart-rate variability (HRV) — a widely used "readiness" or recovery signal — by first staging sleep from body movement and then aggregating HRV over the epochs labeled as sleep. In people with movement-disordered sleep, paradigmatically REM sleep behavior disorder (RBD), which is strongly associated with Parkinson's disease, this fails: dream-enactment movement during REM is misclassified as wakefulness, so valid cardiac data is discarded and the resulting figure is unreliable. This disclosure describes a method that computes the nightly HRV figure over the full in-bed window — without conditioning on the device's sleep/wake or sleep-stage classification — and aggregates it with a robust order statistic (the median) rather than the arithmetic mean. Staging-independence prevents movement from excluding valid data; the robust estimator absorbs the movement artifact the retained periods inject. The result is a stable, staging-independent overnight resting-autonomic figure suited to night-to-night trend tracking on consumer hardware, serving a population that movement-based sleep staging systematically fails. Numerous embodiments are described (robust estimators, HRV metrics, window definitions, sensor types, and confidence gating).
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
ziemski, gerard, "Robust overnight heart-rate-variability estimation independent of sleep/wake staging, for movement-disordered sleep (REM sleep behavior disorder / Parkinson's)", Technical Disclosure Commons, ()
https://www.tdcommons.org/dpubs_series/10443