PPG-Bench: A Unified Benchmark for PPG Foundation Models

Key points

• Photoplethysmography (PPG) is increasingly recorded by wearable and clinical devices at enormous scale, and a growing number of PPG foundation models (FMs) have been developed to compute reusable representations for various downstream tasks.1
• These models are difficult to compare: different evaluations use different datasets, task definitions, preprocessing, downstream heads, hyperparameter tuning, and baselines.1
• We evaluate models under four downstream regimes of increasing adaptability (linear probing, random forests, frozen MLP probes, and end-to-end finetuning), with model-specific hyperparameter optimization and subject-disjoint cross-validation.1
• We also evaluate against three controls: random/constant baseline, a deterministic pulse-rate-variability (PRV) feature set, and a generic time-series FM (MOMENT).1

Sources (1)

• [1]PPG-Bench: A Unified Benchmark for PPG Foundation Models
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 9, 11:34 AM

  We present PPG-Bench, a unifying open-source benchmark evaluating five public PPG FMs across nine open-access datasets and 26 tasks spanning cardiovascular state estimation, activity, sleep, screening, and identification.

  Photoplethysmography (PPG) is increasingly recorded by wearable and clinical devices at enormous scale, and a growing number of PPG foundation models (FMs) have been developed to compute reusable representations for various downstream tasks.

Extractive summary: sentences quoted from the sources.