Objective gait assessment in Parkinson’s Disease (PD) is limited by the absence of large, diverse, and clinically annotated motion datasets. We introduce CARE-PD, the largest publicly available archive of 3D mesh gait data for PD, and the first multi-site collection spanning 9 cohorts from 8 clinical centers. All recordings (RGB video or motion capture) are converted into anonymized SMPL meshes via a harmonized preprocessing pipeline. CARE-PD supports two key benchmarks: supervised clinical score prediction (estimating Unified Parkinson’s Disease Rating Scale, UPDRS, gait scores) and unsupervised motion pretext tasks (2D-to-3D keypoint lifting and full-body 3D reconstruction). Clinical prediction is evaluated under four generalization protocols: within-dataset, cross-dataset, leave-one-dataset-out, and multi-dataset in-domain adaptation. To assess clinical relevance, we compare state-of-the-art motion encoders with a traditional gait-feature baseline, finding that encoders consistently outperform handcrafted features. Pretraining on CARE-PDreduces MPJPE (from 60.8 mm to 7.5 mm) and boosts PD severity macro-F1 by 17 percentage points, underscoring the value of clinically curated, diverse training data. CARE-PD and all benchmark code are released for non-commercial research.
Dataset composition and statistics. The table presents a comprehensive overview of the CARE-PD dataset, showing the distribution of participants, recording sessions, and clinical assessments across the 9 contributing cohorts from 8 international clinical centers. Each cohort's contribution includes detailed demographics, UPDRS scores, and gait recording specifications.
Overview of CARE-PD preprocessing and experimental design. Left: Unified pipeline for extracting SMPL gait meshes from MoCap and video data, followed by model-specific formatting. Right: Benchmarking setup across two pipelines (representation learning vs. gait features), pretext tasks, and four evaluation protocols.
We invite researchers, clinicians, and institutions to contribute to the growing CARE-PD dataset. By joining our collaborative initiative, you can help advance Parkinson's disease research.
We'll work with you to ensure proper data anonymization, ethical compliance, and seamless integration into the CARE-PD pipeline.
Ready to contribute to advancing PD research?
@inproceedings{adeli2025carepd,
title={CARE-PD: A Multi-Site Anonymized Clinical Dataset for Parkinson’s Disease Gait Assessment},
author={Vida Adeli, Ivan Klabučar, Javad Rajabi, Benjamin Filtjens, Soroush Mehraban, Diwei Wang, Hyewon Seo, Trung-Hieu Hoang, Minh N. Do, Candice Muller, Claudia Neves de Oliveira, Daniel Boari Coelho, Pieter Ginis, Moran Gilat, Alice Nieuwboer, Joke Spildooren, J. Lucas McKay, Hyeokhyen Kwon, Gari Clifford, Christine D. Esper, Stewart A. Factor, Imari Genias, Amirhossein Dadashzadeh, Leia Shum, Alan Whone, Majid Mirmehdi, Andrea Iaboni, Babak Taati},
booktitle={NeurIPS},
year={2025}
}If you use the CARE-PD dataset, please cite our paper and all applicable original dataset papers when using the corresponding cohorts.
Important: Each cohort in CARE-PD comes from original research studies. Please ensure you cite all relevant original papers to give proper credit to the data contributors.
Complete list of citations: View all original dataset citations →
Please read carefully the terms and conditions and any accompanying documentation before you download and/or use the CARE-PD dataset.
This dataset is licensed under a Creative Commons
Attribution–NonCommercial–NoDerivatives 4.0 International License (CC BY-NC-ND 4.0).
It may be used for non-commercial research and educational purposes with appropriate citation but may not be modified or redistributed in altered form
(Read the full license).
By using this dataset, you agree to comply with our terms and conditions. Read full terms →