PPG-Distill: Efficient Photoplethysmography Signals Analysis via Foundation Model Distillation
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866911252855390208 |
|---|---|
| author | Ni, Juntong Kataria, Saurabh Tang, Shengpu Yang, Carl Hu, Xiao Jin, Wei |
| author_facet | Ni, Juntong Kataria, Saurabh Tang, Shengpu Yang, Carl Hu, Xiao Jin, Wei |
| contents | Photoplethysmography (PPG) is widely used in wearable health monitoring, yet large PPG foundation models remain difficult to deploy on resource-limited devices. We present PPG-Distill, a knowledge distillation framework that transfers both global and local knowledge through prediction-, feature-, and patch-level distillation. PPG-Distill incorporates morphology distillation to preserve local waveform patterns and rhythm distillation to capture inter-patch temporal structures. On heart rate estimation and atrial fibrillation detection, PPG-Distill improves student performance by up to 21.8% while achieving 7X faster inference and reducing memory usage by 19X, enabling efficient PPG analysis on wearables. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2509_19215 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | PPG-Distill: Efficient Photoplethysmography Signals Analysis via Foundation Model Distillation Ni, Juntong Kataria, Saurabh Tang, Shengpu Yang, Carl Hu, Xiao Jin, Wei Machine Learning Photoplethysmography (PPG) is widely used in wearable health monitoring, yet large PPG foundation models remain difficult to deploy on resource-limited devices. We present PPG-Distill, a knowledge distillation framework that transfers both global and local knowledge through prediction-, feature-, and patch-level distillation. PPG-Distill incorporates morphology distillation to preserve local waveform patterns and rhythm distillation to capture inter-patch temporal structures. On heart rate estimation and atrial fibrillation detection, PPG-Distill improves student performance by up to 21.8% while achieving 7X faster inference and reducing memory usage by 19X, enabling efficient PPG analysis on wearables. |
| title | PPG-Distill: Efficient Photoplethysmography Signals Analysis via Foundation Model Distillation |
| topic | Machine Learning |
| url | https://arxiv.org/abs/2509.19215 |