Wearable Accelerometer Foundation Models for Health via Knowledge Distillation

2024-12-15

arXiv.org

AI Summary

To address the low accuracy of accelerometer-based health monitoring in wearable devices, this paper introduces the first accelerometer foundation model tailored for health diagnostics. Methodologically, we propose a novel cross-modal knowledge distillation framework (PPG → accelerometer), integrated with self-supervised pretraining and cross-modal embedding alignment, trained on 20 million minutes of unlabeled real-world wearable data. Our core contribution is the first demonstration that low-power accelerometers can acquire high-fidelity, PPG-level health representations—establishing a new paradigm wherein accelerometers serve as universal health foundation models. Experiments show 23–49% reductions in heart rate and heart rate variability prediction errors, a 99.2% Top-1 accuracy in PPG–accelerometer cross-modal embedding retrieval, and significantly improved generalization performance across multiple downstream physiological biomarkers.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Cognitive Modeling & Cognitive Systems: Other Foundations of Cognitive Modeling & Systems
• Computer Vision: Other Foundations of Computer Vision

Application Category

• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Search and Retrieval-Augmented AI: Personalized, context-aware and across-device search

Abstract

Modern wearable devices can conveniently record various biosignals in the many different environments of daily living, enabling a rich view of individual health. However, not all biosignals are the same: high-fidelity biosignals, such as photoplethysmogram (PPG), contain more physiological information, but require optical sensors with a high power footprint. Alternatively, a lower-fidelity biosignal such as accelerometry has a significantly smaller power footprint and is available in almost any wearable device. While accelerometry is widely used for activity recognition and fitness, it is less explored for health biomarkers and diagnosis. Here, we show that an accelerometry foundation model can predict a wide variety of health targets. To achieve improved performance, we distill representational knowledge from PPG encoders to accelerometery encoders using 20 million minutes of unlabeled data, collected from ~172K participants in the Apple Heart and Movement Study under informed consent. We observe strong cross-modal alignment on unseen data, e.g., 99.2% top-1 accuracy for retrieving PPG embeddings from accelerometry embeddings. We show that distilled accelerometry encoders have significantly more informative representations compared to self-supervised or supervised encoders trained directly on accelerometry data, observed by at least 23%-49% improved performance for predicting heart rate and heart rate variability. We also show that distilled accelerometry encoders are readily predictive of a wide array of downstream health targets, i.e., they are generalist foundation models. We believe accelerometry foundation models for health may unlock new opportunities for developing digital biomarkers from any wearable device.

Problem

• Accelerometer
• Health Monitoring
• Heart Rate Prediction

Innovation

• Accelerometer
• Machine Learning
• Health Monitoring