Profiling Postural Instability in Parkinson's disease: Multi-Parametric Phenotyping and Clinical Correlates from Markerless RGB-D Tracking

This repository contains the dataset supporting the study "Profiling Postural Instability in Parkinson’s disease: Multi-Parametric Phenotyping and Clinical Correlates from Markerless RGB-D Tracking " (in submission) The dataset includes anonymized clinical and kinematic data from 40 moderate-to-advanced Parkinson's Disease (PD) patients. Data were collected to investigate postural instability and sensory reweighting strategies using a non-invasive, markerless RGB-D sensor (Microsoft Azure Kinect) and Human Pose Estimation (HPE). Patients performed a quiet stance task under two visual conditions: Eyes Open (EO) and Eyes Closed (EC). The dataset provides stabilometric parameters extracted from the 3D Center of Mass (CoM) trajectories along the medio-lateral (ML), antero-posterior (AP), and vertical (UD) axes. Please README.txt for a detailed explanation of the dataset contents. For full methodological details, mathematical definitions, and study conclusions, please refer to the associated manuscript.