Guardado en:
| Autores principales: | , , , , , , , , , , , , , , , , , , , , , , , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2025
|
| Materias: | |
| Acceso en línea: | https://arxiv.org/abs/2510.04312 |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866909825812660224 |
|---|---|
| author | Adeli, Vida Klabucar, Ivan Rajabi, Javad Filtjens, Benjamin Mehraban, Soroush Wang, Diwei Seo, Hyewon Hoang, Trung-Hieu Do, Minh N. Muller, Candice Oliveira, Claudia Coelho, Daniel Boari Ginis, Pieter Gilat, Moran Nieuwboer, Alice Spildooren, Joke Mckay, Lucas Kwon, Hyeokhyen Clifford, Gari Esper, Christine Factor, Stewart Genias, Imari Dadashzadeh, Amirhossein Shum, Leia Whone, Alan Mirmehdi, Majid Iaboni, Andrea Taati, Babak |
| author_facet | Adeli, Vida Klabucar, Ivan Rajabi, Javad Filtjens, Benjamin Mehraban, Soroush Wang, Diwei Seo, Hyewon Hoang, Trung-Hieu Do, Minh N. Muller, Candice Oliveira, Claudia Coelho, Daniel Boari Ginis, Pieter Gilat, Moran Nieuwboer, Alice Spildooren, Joke Mckay, Lucas Kwon, Hyeokhyen Clifford, Gari Esper, Christine Factor, Stewart Genias, Imari Dadashzadeh, Amirhossein Shum, Leia Whone, Alan Mirmehdi, Majid Iaboni, Andrea Taati, Babak |
| contents | 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-PD reduces MPJPE (from 60.8mm to 7.5mm) 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 at https://neurips2025.care-pd.ca/. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2510_04312 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | CARE-PD: A Multi-Site Anonymized Clinical Dataset for Parkinson's Disease Gait Assessment Adeli, Vida Klabucar, Ivan Rajabi, Javad Filtjens, Benjamin Mehraban, Soroush Wang, Diwei Seo, Hyewon Hoang, Trung-Hieu Do, Minh N. Muller, Candice Oliveira, Claudia Coelho, Daniel Boari Ginis, Pieter Gilat, Moran Nieuwboer, Alice Spildooren, Joke Mckay, Lucas Kwon, Hyeokhyen Clifford, Gari Esper, Christine Factor, Stewart Genias, Imari Dadashzadeh, Amirhossein Shum, Leia Whone, Alan Mirmehdi, Majid Iaboni, Andrea Taati, Babak Computer Vision and Pattern Recognition 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-PD reduces MPJPE (from 60.8mm to 7.5mm) 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 at https://neurips2025.care-pd.ca/. |
| title | CARE-PD: A Multi-Site Anonymized Clinical Dataset for Parkinson's Disease Gait Assessment |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2510.04312 |