PaperarXiv:2412.17860

EnhancePPG: Improving PPG-based Heart Rate Estimation with Self-Supervision and Augmentation

PreprintDated Dec 20, 2024Added Nov 17, 2026

  • B
  • B
  • B
  • K
  • W
  • B
  • P
  • M
  • +1

Authors: Benfenati, Belloni, Burrello, Kasnesis, Wang, Benini, Poncino, Macii, Jahier Pagliari

arXiv preprint 2412.17860

TL;DR

U-Net-style self-supervised reconstruction pre-training plus data augmentation on a small edge HR model (modified PULSE). PPG-DaLiA MAE falls from 4.03 to 3.54 bpm with about the same inference cost. No code or weights found.

  • Form factor: wrist

Abstract

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

Verbatim from arXiv (opens in a new tab) · arXiv metadata (abstracts are released under CC0)

Results on ChatPPG leaderboards · 1

All leaderboards
LeaderboardMethodResultRank
PPG-DaLiA MAE (BPM)Heart rate estimationEnhancePPG3.54MAE (BPM), lower is better#3 of 20

1 row is credited to this paper. Ranks apply within each leaderboard only, as reported by the source.

Tasks & signals

Tasks

Signals

Keywords

Heart RateSelf SupervisedData AugmentationPULSETCNEdgePPG DaLiAPolito

Datasets used

How it works

How Method works: Setup → Method → TargetSetup: Signals (ppg, acc); Form factor (wrist). Method: Method. Target: Heart rate.SetupSignals: ppg, accSignalsppg, accForm factor: wristForm factorwristMethodMethodTargetHeart rateHeart rate

Cite this paper

Cite this paper with the fields recorded here. Check it against the publisher before submitting.

EnhancePPG: Improving PPG-based Heart Rate Estimation with Self-Supervision and Augmentation. (2024). arXiv preprint 2412.17860. https://arxiv.org/abs/2412.17860
@misc{benfenatienhance2024enhanceppg,
  title         = {{EnhancePPG: Improving PPG-based Heart Rate Estimation with Self-Supervision and Augmentation}},
  year          = {2024},
  howpublished  = {arXiv preprint 2412.17860},
  eprint        = {2412.17860},
  archivePrefix = {arXiv}
}

Source metadata: 2024

Record updated Nov 17, 2026

Added to the ChatPPG research hub