Comprehensive heart rate variability estimation in relation to sleep state architecture: a retrospective observational cohort study on Apple Watch heart rate data

This dataset accompanies the manuscript " Comprehensive HRV estimation pipeline in Python using Neurokit2" under review in MethodsX and released under https://doi.org/10.5281/zenodo.5736571. The presented method is an adaptation of NeuroKit2 to simplify and automate the computation of the various mathematical estimates of heart rate variability (HRV) or similar time series. The dataset contains the Jupyter notebook and computed datasets resulting from the application of the method to an Apple Watch polysomnography dataset published elsewhere (see references). The methods are described for computation of 124 HRV measures including measures with a dynamic, time-series-specific optimal time delay-based complexity estimation with a user-definable time window length. As part of the test application of the methodology, I present an approach to studying the dynamic relationships between sleep state architecture and multi-dimensional HRV metrics in 31 subjects. I gratefully acknowledge the excellent support from the NeuroKit team.