EnhancePPG: Improving PPG-based Heart Rate Estimation with Self-Supervision and Augmentation
This work addresses the need for accurate health monitoring in wearables with an incremental improvement over existing methods.
The paper tackles the problem of heart rate estimation from PPG signals in wearable devices by integrating self-supervised learning with data augmentation, resulting in a 12.2% improvement in error reduction from 4.03 BPM to 3.54 BPM on the PPG-DaLiA dataset.
Heart rate (HR) estimation from photoplethysmography (PPG) signals is a key feature of modern wearable devices for health and wellness monitoring. While deep learning models show promise, their performance relies on the availability of large datasets. We present EnhancePPG, a method that enhances state-of-the-art models by integrating self-supervised learning with data augmentation (DA). Our approach combines self-supervised pre-training with DA, allowing the model to learn more generalizable features, without needing more labelled data. Inspired by a U-Net-like autoencoder architecture, we utilize unsupervised PPG signal reconstruction, taking advantage of large amounts of unlabeled data during the pre-training phase combined with data augmentation, to improve state-of-the-art models' performance. Thanks to our approach and minimal modification to the state-of-the-art model, we improve the best HR estimation by 12.2%, lowering from 4.03 Beats-Per-Minute (BPM) to 3.54 BPM the error on PPG-DaLiA. Importantly, our EnhancePPG approach focuses exclusively on the training of the selected deep learning model, without significantly increasing its inference latency