HRV Feature Dataset for Cross-Dataset Machine Learning Screening of REM Sleep Behavior Disorder

This dataset contains a comprehensive collection of Heart Rate Variability (HRV)\u2013derived features extracted from four independent cohorts involved in studies on REM Sleep Behavior Disorder (RBD) and healthy sleep. The data are provided as a single harmonized CSV file, enabling direct use in machine learning, statistical modeling, and cross-cohort validation studies focusing on autonomic function during sleep.The dataset includes all HRV feature domains typically used in sleep and neurodegeneration research, including:Time-domain metrics (e.g., SDNN, RMSSD, pNN50)Frequency-domain metrics (e.g., LF, HF, LF\/HF ratio)Nonlinear dynamics measures (e.g., Sample Entropy, Detrended Fluctuation Analysis, Poincar\u00e9 indices)Geometrical and morphological descriptors of RR interval distribution.The dataset supports research on wearable-enabled sleep monitoring, early detection of synucleinopathies, and generalization of machine learning models across heterogeneous sleep cohorts. It is particularly suited for benchmarking cross-dataset and transfer learning approaches for RBD detection.By consolidating HRV features from multiple sources into a unified and openly accessible structure, this resource enables reproducibility, accelerates methodological comparisons, and facilitates feasibility assessments of large-scale screening pipelines for neurodegenerative risk monitoring based on minimal and non-invasive sensing.