Care-PD
收藏DataCite Commons2026-05-08 更新2025-06-14 收录
下载链接：
https://borealisdata.ca/citation?persistentId=doi:10.5683/SP3/TWIKMK
下载链接
链接失效反馈官方服务：
资源简介：
<!DOCTYPE html>
<html lang="en">
<body>
<h1>Overview</h1>
<h3> Please read carefully the terms and conditions and any accompanying documentation at <a href="https://neurips2025.care-pd.ca/terms-of-use.html"
target="_blank"
style="color:#0066cc; text-decoration:underline;">
neurips2025.care-pd.ca/terms-of-use
</a> before you download and/or use the CARE-PD dataset.
</h3>
<p>
Project page: <a href="https://neurips2025.care-pd.ca/"
target="_blank"
style="color:#0066cc; text-decoration:underline;">
https://neurips2025.care-pd.ca/
</a>
</p>
<p>
CARE-PD is the largest publicly available archive of 3D mesh gait data for Parkinson's Disease (PD) and the first to include data collected across multiple sites.
The dataset aggregates 9 cohorts from 8 clinical sites, including 362 participants spanning a range of disease severity.
All recordings—whether from RGB video or motion capture—are unified into anonymized SMPL body gait meshes through a curated harmonization pipeline.
</p>
<h3>This dataset enables two main benchmarks:</h3>
<ol>
<li>Supervised clinical score prediction: Estimating UPDRS gait scores from 3D meshes</li>
<li>Unsupervised motion pretext tasks for Parkinsonian gait representation learning</li>
</ol>
<h2>Dataset Contents</h2>
<p>CARE-PD consists of 9 harmonized datasets:</p>
<ol>
<li>3DGait&nbsp;– Clinical gait recordings with UPDRS scores</li>
<li>BMCLab&nbsp;– Gait recordings with medication status and UPDRS scores (original license: CC BY 4.0)</li>
<li>DNE&nbsp;– Contains healthy, Parkinson's, and other neurological conditions (original license: CC BY 4.0)</li>
<li>E-LC&nbsp;– Medication status (on/off) and PD subtypes</li>
<li>KUL-DT-T&nbsp;– Freezer/non-freezer subtypes</li>
<li>PD-GaM&nbsp;– Clinical gait recordings with UPDRS scores</li>
<li>T-SDU&nbsp;– Ambient walking recordings</li>
<li>T-SDU-PD&nbsp;– PD patient walking with UPDRS scores</li>
<li>T-LTC&nbsp;– Ambient walking recordings</li>
</ol>
<h3>Canonicalized SMPL files</h3>
<p>
<code>*_canonical.pkl</code> files in the
<code>Canonicalized_SMPL_pickles</code> folder keep the same nested dataset
format as the original pickles. The canonical versions change only the motion coordinates:
</p>
<p>
<code>pose/trans</code> are rotated so the motion uses a shared coordinate system.
</p>
<pre><code>x = lateral
y = up
z = forward</code></pre>
<p>They also preprocess translation so:</p>
<ul>
<li>The first frame starts at <code>x=0, z=0</code></li>
<li>The body stands/walks on <code>y=0</code></li>
</ul>
<p>
<b>Note:</b> <code>KUL-DT-T</code> and <code>E-LC</code> are the only datasets that are not purely straight walking sequences, so for these two datasets the subject is canonicalized to start facing <code>z+</code> in the first frame.
</p>
<h2>Data Structure</h2>
<p>The main SMPL datasets are provided in a standardized format:</p>
<pre><code>{
"anonymized_subject_id": {
"anonymized_walk_id": {
"pose": array, # SMPL pose parameters (shape varies by dataset)
"trans": array, # Translation data
"beta": array, # Body shape parameters (zeros for privacy)
"fps": int, # Frames per second (standardized)
"UPDRS_GAIT": int, # Clinical score (0-3) or None if unavailable
"medication": str, # Medication status or None if unavailable
"other": str # Additional labels or None if unavailable
}
}
}</code></pre>
<p>Additionally, we provide h36m, HumanML3D, and SMPL_6D formats.</p>
<h2>Getting Started</h2>
<p>Please refer to <a href="https://github.com/TaatiTeam/CARE-PD">https://github.com/TaatiTeam/CARE-PD</a> for getting started with the dataset.</p>
<h2>Benchmarks</h2>
<p>CARE-PD includes data splits to test generalization:</p>
<ol>
<li>6-Fold (split per subject)</li>
<li>Leave-one-subject-out</li>
<li>Fixed train-test splits (split per subject)</li>
</ol>
<p>The former two are only provided for the supervised clinical score prediction task.</p>
<h2>Terms of Use</h2>
<p>
By accessing and using this database (the "Database"), users ("Users") acknowledge and agree to comply with the following conditions:
</p>
<ol>
<li><strong>License and Attribution</strong>
<ul>
<li>The Database is publicly released under a Creative Commons Attribution-NonCommercial (CC&nbsp;BY-NC&nbsp;4.0) license.</li>
<li>Users must provide appropriate attribution by citing the Database and the original publications associated with each dataset accessed from the Database.</li>
</ul>
</li>
<li><strong>Data Privacy and Ethics</strong>
<ul>
<li>Users must not attempt to identify, contact, or otherwise compromise the anonymity of any individuals whose data is included in the Database.</li>
<li>All use of the data must comply with applicable ethical guidelines and legal regulations, including privacy laws (e.g., GDPR, HIPAA, PIPEDA).</li>
</ul>
</li>
<li><strong>Data Handling and Security</strong>
<ul>
<li>Users must maintain appropriate data security measures to prevent unauthorized access, sharing, or use of the data.</li>
<li>Users are encouraged, but not required, to direct third parties to the original Database URL rather than re-hosting the data.</li>
</ul>
</li>
<li><strong>Intellectual Property Notice</strong>
<ul>
<li>Copyright and other rights remain with the original data providers.</li>
</ul>
</li>
<li><strong>Disclaimer of Warranty</strong>
<ul>
<li>Use of the Database is subject to Section&nbsp;5 (Disclaimer of Warranties and Limitation of Liability) of the CC&nbsp;BY-NC&nbsp;4.0 licence.</li>
</ul>
</li>
</ol>
<p>By using the Database, Users expressly acknowledge and agree to abide by these Terms of Use.</p>
<h2>Citation</h2>
<p>If you use CARE-PD in your research, please cite:</p>
<p>Adeli V, Klabučar I, Rajabi J, Filtjens B, Mehraban S, Wang D, Seo H, Hoang T-H, Do MN, Muller C, Neves de Oliveira C, Boari Coelho D, Ginis P, Gilat M, Nieuwboer A, Spildooren J, McKay JL, Kwon H, Clifford G, Esper CD, Factor SA, Genias I, Dadashzadeh A, Shum L, Whone A, Mirmehdi M, Iaboni A, Taati B.
CARE-PD: A Multi-Site Anonymized Clinical Dataset for Parkinson’s Disease Gait Assessment.
In: Advances in Neural Information Processing Systems (NeurIPS); 2025.</p>
<p>Additionally, please cite the relevant datasets:</p>
<ol>
<li>3DGait
<ul>
<li>Diwei Wang, Chaima Zouaoui, Jinhyeok Jang, Hassen Drira, and Hyewon Seo. 2023. Video-Based Gait Analysis for Assessing Alzheimer’s Disease and Dementia with Lewy Bodies. In <em>Applications of Medical Artificial Intelligence: Second International Workshop, AMAI&nbsp;2023, Held in Conjunction with MICCAI&nbsp;2023</em>, Vancouver, BC, Canada, October&nbsp;8, 2023, Proceedings. Springer-Verlag, Berlin, Heidelberg, 72–82. https://doi.org/10.1007/978-3-031-47076-9_8</li>
</ul>
</li>
<li>BMCLab
<ul>
<li>Shida TKF, Costa TM, de Oliveira CEN, de Castro Treza R, Hondo SM, Los Angeles E, Bernardo C, Dos Santos de Oliveira L, de Jesus Carvalho M, Coelho DB. A public data set of walking full-body kinematics and kinetics in individuals with Parkinson's disease. <em>Front Neurosci.</em> 2023 Feb&nbsp;16;17:992585. doi: 10.3389/fnins.2023.992585. PMID: 36875659; PMCID: PMC9978741.</li>
</ul>
</li>
<li>DNE
<ul>
<li>Hoang TH, Zallek C, Do MN. Smartphone-Based Digitized Neurological Examination Toolbox for Multi-test Neurological Abnormality Detection and Documentation. <em>IEEE J Biomed Health Inform.</em> 2024 Aug&nbsp;26;PP. doi: 10.1109/JBHI.2024.3439492. Epub ahead of print. PMID: 39186431.</li>
<li>Hoang TH, Zehni M, Xu H, Heintz G, Zallek C, Do MN. Towards a Comprehensive Solution for a Vision-Based Digitized Neurological Examination. <em>IEEE J Biomed Health Inform.</em> 2022 Aug&nbsp;26(8):4020-4031. doi: 10.1109/JBHI.2022.3167927. Epub 2022 Aug 11. PMID: 35439148; PMCID: PMC9707344.</li>
</ul>
</li>
<li>E-LC
<ul>
<li>Lucas McKay J, Goldstein FC, Sommerfeld B, Bernhard D, Perez Parra S, Factor SA. Freezing of Gait can persist after an acute levodopa challenge in Parkinson's disease. <em>NPJ Parkinsons Dis.</em> 2019 Nov&nbsp;22;5:25. doi: 10.1038/s41531-019-0099-z. PMID: 31799377; PMCID: PMC6874572.</li>
<li>Kwon H, Clifford GD, Genias I, Bernhard D, Esper CD, Factor SA, McKay JL. An Explainable Spatial-Temporal Graphical Convolutional Network to Score Freezing of Gait in Parkinsonian Patients. <em>Sensors (Basel).</em> 2023 Feb&nbsp;4;23(4):1766. doi: 10.3390/s23041766. PMID: 36850363; PMCID: PMC9968199.</li>
</ul>
</li>
<li>KUL-DT-T
<ul>
<li>Spildooren J, Vercruysse S, Desloovere K, Vandenberghe W, Kerckhofs E, Nieuwboer A. Freezing of gait in Parkinson's disease: the impact of dual-tasking and turning. <em>Mov Disord.</em> 2010 Nov&nbsp;15;25(15):2563–70. doi: 10.1002/mds.23327. PMID: 20632376.</li>
<li>Filtjens B, Ginis P, Nieuwboer A, Slaets P, Vanrumste B. Automated freezing of gait assessment with marker-based motion capture and multi-stage spatial-temporal graph convolutional neural networks. <em>J Neuroeng Rehabil.</em> 2022 May&nbsp;21;19(1):48. doi: 10.1186/s12984-022-01025-3. PMID: 35597950; PMCID: PMC9124420.</li>
</ul>
</li>
<li>PD-GaM
<ul>
<li>Vida Adeli, Soroush Mehraban, Majid Mirmehdi, Alan Whone, Benjamin Filtjens, Amirhossein Dadashzadeh, Alfonso Fasano, and Andrea Iaboni, Babak Taati. GAITGen: Disentangled motion-pathology impaired gait generative model—bringing motion generation to the clinical domain. <em>arXiv</em> preprint arXiv:2503.22397, 2025.</li>
<li>Amirhossein Dadashzadeh, Shuchao Duan, Alan Whone, and Majid Mirmehdi. PeCop: Parameter efficient continual pretraining for action quality assessment. In <em>Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision</em>, 42–52, 2024.</li>
</ul>
</li>
<li>T-SDU
<ul>
<li>Adeli V, Klabučar I, Rajabi J, Filtjens B, Mehraban S, Wang D, Seo H, Hoang T-H, Do MN, Muller C, Neves de Oliveira C, Boari Coelho D, Ginis P, Gilat M, Nieuwboer A, Spildooren J, McKay JL, Kwon H, Clifford G, Esper CD, Factor SA, Genias I, Dadashzadeh A, Shum L, Whone A, Mirmehdi M, Iaboni A, Taati B.
CARE-PD: A Multi-Site Anonymized Clinical Dataset for Parkinson’s Disease Gait Assessment.
In: Advances in Neural Information Processing Systems (NeurIPS); 2025.</li>
</ul>
</li>
<li>T-SDU-PD
<ul>
<li>Adeli V, Klabučar I, Rajabi J, Filtjens B, Mehraban S, Wang D, Seo H, Hoang T-H, Do MN, Muller C, Neves de Oliveira C, Boari Coelho D, Ginis P, Gilat M, Nieuwboer A, Spildooren J, McKay JL, Kwon H, Clifford G, Esper CD, Factor SA, Genias I, Dadashzadeh A, Shum L, Whone A, Mirmehdi M, Iaboni A, Taati B.
CARE-PD: A Multi-Site Anonymized Clinical Dataset for Parkinson’s Disease Gait Assessment.
In: Advances in Neural Information Processing Systems (NeurIPS); 2025.</li>
</ul>
</li>
<li>T-LTC
<ul>
<li>Adeli V, Klabučar I, Rajabi J, Filtjens B, Mehraban S, Wang D, Seo H, Hoang T-H, Do MN, Muller C, Neves de Oliveira C, Boari Coelho D, Ginis P, Gilat M, Nieuwboer A, Spildooren J, McKay JL, Kwon H, Clifford G, Esper CD, Factor SA, Genias I, Dadashzadeh A, Shum L, Whone A, Mirmehdi M, Iaboni A, Taati B.
CARE-PD: A Multi-Site Anonymized Clinical Dataset for Parkinson’s Disease Gait Assessment.
In: Advances in Neural Information Processing Systems (NeurIPS); 2025.</li>
</ul>
</li>
</ol>
<h2>Acknowledgments</h2>
<p>We thank all participating research institutions and subjects who made this dataset possible.</p>
</body>
</html>
# 概述
## 重要提示
请仔细阅读<a href="https://neurips2025.care-pd.ca/terms-of-use.html">neurips2025.care-pd.ca/terms-of-use</a>中的条款与条件及随附文档，再下载或使用CARE-PD数据集。
项目页面：<a href="https://neurips2025.care-pd.ca/">https://neurips2025.care-pd.ca/</a>
CARE-PD是目前规模最大的公开帕金森病（Parkinson's Disease, PD）三维网格步态数据档案库，也是首个纳入多中心采集数据的同类数据集。该数据集整合了来自8个临床站点的9个队列数据，涵盖362名不同疾病严重程度的受试者。所有录制数据——无论是来自RGB视频还是动作捕捉——均通过标准化的统一处理管线，转换为匿名化的SMPL人体步态网格。
本数据集可支撑两大核心基准任务：
1. 监督式临床评分预测：基于三维网格估算统一帕金森病评定量表（Unified Parkinson's Disease Rating Scale, UPDRS）步态评分
2. 帕金森步态表征学习的无监督运动预训练任务
## 数据集内容
CARE-PD包含9个经过统一标准化处理的数据集：
1. 3DGait：携带UPDRS评分的临床步态录制数据
2. BMCLab：携带用药状态与UPDRS评分的步态录制数据（原始授权协议：CC BY 4.0）
3. DNE：涵盖健康人群、帕金森病患者及其他神经系统疾病患者数据（原始授权协议：CC BY 4.0）
4. E-LC：包含用药状态（服药/未服药）及帕金森病亚型信息
5. KUL-DT-T：涵盖步态冻结/非冻结亚型数据
6. PD-GaM：携带UPDRS评分的临床步态录制数据
7. T-SDU：自然行走录制数据
8. T-SDU-PD：携带UPDRS评分的帕金森病患者行走数据
9. T-LTC：自然行走录制数据
## 数据结构
核心SMPL数据集采用标准化格式存储，结构如下：
json
{
"anonymized_subject_id": {
"anonymized_walk_id": {
"pose": 数组, # SMPL姿态参数（维度因数据集而异）
"trans": 数组, # 人体平移数据
"beta": 数组, # 人体体型参数（为保护隐私设为全零数组）
"fps": 整数, # 帧率（已统一标准化）
"UPDRS_GAIT": 整数, # 临床步态评分（取值范围0-3，不可用则为None）
"medication": 字符串, # 用药状态，不可用则为None
"other": 字符串 # 额外标注信息，不可用则为None
}
}
}
此外，本数据集还提供h36m、HumanML3D及SMPL_6D三种额外格式的数据。
## 快速入门
请访问<a href="https://github.com/TaatiTeam/CARE-PD">https://github.com/TaatiTeam/CARE-PD</a>获取数据集的使用指南与快速入门教程。
## 基准任务
CARE-PD提供了多种数据集划分方式，用于测试模型的泛化能力：
1. 6折交叉验证（按受试者个体划分）
2. 留一受试者法
3. 固定训练-测试集划分（按受试者个体划分）
其中前两种划分方式仅适用于监督式临床评分预测任务。
## 使用条款
用户访问或使用本数据库（下称“数据库”）即表示其知晓并同意遵守以下条款：
1. **授权与署名**
- 本数据库采用知识共享署名-非商业性使用（CC BY-NC 4.0）协议公开发布。
- 用户需对本数据库及所使用的各原始数据集的相关学术文献进行恰当引用，以完成学术署名。
2. **数据隐私与伦理**
- 用户不得尝试识别、联系或损害数据库中所有受试者的匿名性。
- 所有数据使用行为必须符合适用的伦理准则与法律法规，包括但不限于GDPR、HIPAA、PIPEDA等隐私保护相关法律。
3. **数据处理与安全**
- 用户需采取适当的数据安全措施，防止未经授权的访问、共享或使用数据集。
- 鼓励（但非强制要求）用户将第三方引导至本数据库的原始链接，而非自行重新托管数据集副本。
4. **知识产权声明**
- 本数据集的版权及其他相关知识产权仍归原始数据提供方所有。
5. **免责声明**
- 本数据库的使用需遵循CC BY-NC 4.0协议第5条（免责声明与责任限制）的相关规定。
使用本数据库即表示用户明确知晓并同意遵守本使用条款。
## 引用方式
若您在研究中使用CARE-PD数据集，请引用以下文献：
Adeli V, Klabučar I, Rajabi J, Filtjens B, Mehraban S, Wang D, Seo H, Hoang T-H, Do MN, Muller C, Neves de Oliveira C, Boari Coelho D, Ginis P, Gilat M, Nieuwboer A, Spildooren J, McKay JL, Kwon H, Clifford G, Esper CD, Factor SA, Genias I, Dadashzadeh A, Shum L, Whone A, Mirmehdi M, Iaboni A, Taati B.
CARE-PD: A Multi-Site Anonymized Clinical Dataset for Parkinson’s Disease Gait Assessment.
In: Advances in Neural Information Processing Systems (NeurIPS); 2025.
此外，请引用所使用数据集的相关学术文献：
1. 3DGait
- Diwei Wang, Chaima Zouaoui, Jinhyeok Jang, Hassen Drira, Hyewon Seo. 2023. 基于视频的步态分析用于评估阿尔茨海默病与路易体痴呆. 见：《医学人工智能应用：第二届国际研讨会，AMAI 2023，与MICCAI 2023联合举办》，加拿大不列颠哥伦比亚省温哥华，2023年10月8日，会议论文集. Springer-Verlag, 柏林, 海德堡, 72–82. https://doi.org/10.1007/978-3-031-47076-9_8
2. BMCLab
- Shida TKF, Costa TM, de Oliveira CEN, de Castro Treza R, Hondo SM, Los Angeles E, Bernardo C, Dos Santos de Oliveira L, de Jesus Carvalho M, Coelho DB. 帕金森病患者全身行走运动学与动力学公开数据集. 《Front Neurosci.》（《神经科学前沿》）. 2023 Feb 16;17:992585. doi: 10.3389/fnins.2023.992585. PMID: 36875659; PMCID: PMC9978741.
3. DNE
- Hoang TH, Zallek C, Do MN. 基于智能手机的数字化神经系统检查工具箱：用于多检测异常的检测与记录. 《IEEE J Biomed Health Inform.》（《IEEE生物医学与健康信息学杂志》）. 2024 Aug 26;PP. doi: 10.1109/JBHI.2024.3439492. Epub ahead of print. PMID: 39186431.
- Hoang TH, Zehni M, Xu H, Heintz G, Zallek C, Do MN. 面向视觉数字化神经系统检查的综合解决方案. 《IEEE J Biomed Health Inform.》（《IEEE生物医学与健康信息学杂志》）. 2022 Aug 26(8):4020-4031. doi: 10.1109/JBHI.2022.3167927. Epub 2022 Aug 11. PMID: 35439148; PMCID: PMC9707344.
4. E-LC
- Lucas McKay J, Goldstein FC, Sommerfeld B, Bernhard D, Perez Parra S, Factor SA. 帕金森病患者急性左旋多巴挑战后步态冻结仍可持续存在. 《NPJ Parkinsons Dis.》（《npj帕金森病》）. 2019 Nov 22;5:25. doi: 10.1038/s41531-019-0099-z. PMID: 31799377; PMCID: PMC6874572.
- Kwon H, Clifford GD, Genias I, Bernhard D, Esper CD, Factor SA, McKay JL. 可解释时空图卷积网络用于帕金森患者步态冻结评分. 《Sensors (Basel).》（《传感器（巴塞尔）》）. 2023 Feb 4;23(4):1766. doi: 10.3390/s23041766. PMID: 36850363; PMCID: PMC9968199.
5. KUL-DT-T
- Spildooren J, Vercruysse S, Desloovere K, Vandenberghe W, Kerckhofs E, Nieuwboer A. 帕金森病患者的步态冻结：双任务与转身的影响. 《Mov Disord.》（《运动障碍杂志》）. 2010 Nov 15;25(15):2563–70. doi: 10.1002/mds.23327. PMID: 20632376.
- Filtjens B, Ginis P, Nieuwboer A, Slaets P, Vanrumste B. 基于标记动作捕捉与多阶段时空图卷积神经网络的自动化步态冻结评估. 《J Neuroeng Rehabil.》（《神经工程与康复学报》）. 2022 May 21;19(1):48. doi: 10.1186/s12984-022-01025-3. PMID: 35597950; PMCID: PMC9124420.
6. PD-GaM
- Vida Adeli, Soroush Mehraban, Majid Mirmehdi, Alan Whone, Benjamin Filtjens, Amirhossein Dadashzadeh, Alfonso Fasano, Andrea Iaboni, Babak Taati. GAITGen：解耦运动-病理受损步态生成模型——将运动生成引入临床领域. arXiv预印本arXiv:2503.22397, 2025.
- Amirhossein Dadashzadeh, Shuchao Duan, Alan Whone, Majid Mirmehdi. PeCop：用于动作质量评估的参数高效持续预训练. 见：《Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision》（《IEEE/CVF冬季计算机视觉应用会议论文集》）, 42–52, 2024.
7. T-SDU
- Adeli V等. CARE-PD：面向帕金森病步态评估的多中心匿名化临床数据集. 见：神经信息处理系统大会（NeurIPS）进展; 2025.
8. T-SDU-PD
- Adeli V等. CARE-PD：面向帕金森病步态评估的多中心匿名化临床数据集. 见：神经信息处理系统大会（NeurIPS）进展; 2025.
9. T-LTC
- Adeli V等. CARE-PD：面向帕金森病步态评估的多中心匿名化临床数据集. 见：神经信息处理系统大会（NeurIPS）进展; 2025.
## 致谢
我们感谢所有参与本数据集构建的研究机构与受试者。
提供机构：
Borealis
创建时间：
2025-05-13
搜集汇总
数据集介绍
背景与挑战
背景概述
Care-PD是目前最大的公开帕金森病3D步态数据集，首次整合了多中心临床数据，包含362名参与者的匿名化步态信息，支持UPDRS步态评分预测和运动表示学习等任务。数据集采用标准化SMPL格式，并提供了多种数据分割方案以测试模型的泛化能力。
以上内容由遇见数据集搜集并总结生成
