ChatPPG.
Join free
Model

QUMPHY D1 PPG benchmark checkpoints (BP + AF; ResNet18 image / TCN / MLP)

code release

Datasets
PulseDB v2.0, Stanford wrist PPG AF (DeepBeat)
Year
2025
Form factor
Any
PPGOpen codeWeights: open
Code Cite
Figure or first page from the QUMPHY D1 PPG benchmark checkpoints (BP + AF; ResNet18 image / TCN / MLP) source
UnconfirmedCatalog paper date: 2025Added to atlas: Oct 21, 2026

Activity counts measure attention, not model quality. Missing counts are unknown. Publication status follows the linked source; a listed venue alone does not confirm peer review.

QUMPHY D1 releases code and trained checkpoints for a PPG benchmark covering blood-pressure and atrial-fibrillation tasks with ResNet18 image models, temporal convolutional networks, and feature MLPs. The accompanying benchmark paper is in Biomedical Signal Processing and Control / arXiv 2502.19949 under the EU QUMPHY project. GitLab: gitlab.com/qumphy/d1-code.

Signals
PPG
Form factor
Any
License
None declared (all rights reserved by default)
Year
2025
Architecture

Task models from the benchmark: ResNet18 on image representations, TCN (+MLP) on raw time series, MLP on features (checkpoint file names); full model zoo in paper

At a glance
Type
Task model
Modalities
PPG
Form factor
Any
Parameters
Not reported
Architecture
Task models from the benchmark: ResNet18 on image representations, TCN (+MLP) on raw time series, MLP on features (checkpoint file names); full model zoo in paper
Venue
Not reported
Year
2025
arXiv
Not reported
DOI
Not reported
Code
Open (opens in a new tab)
Weights
Open
Licence
None declared (all rights reserved by default)
Caveats
Not reported
Datasets
Added
Oct 21, 2026
Updated
Oct 26, 2026

How it works

How QUMPHY D1 PPG benchmark checkpoints (BP + AF; ResNet18 image / TCN / MLP) works: Input signals → Model → Downstream tasksInput signals: PPG (Photoplethysmogram). Datasets: PulseDB v2.0, Stanford wrist PPG AF (DeepBeat). Model: QUMPHY D1 PPG benchmark checkpoints (BP + AF; ResNet18 image / TCN / MLP) (Convolutional network). Downstream tasks: Blood pressure; AF / rhythm.Input signalsPPG: PhotoplethysmogramPPGPhotoplethysmogramDatasets: PulseDB v2.0, Stanford wrist PPG AF (DeepBeat)DatasetsPulseDB v2.0, Stanford wrist PPG AF…ModelQUMPHY D1 PPG benchmark…Convolutional networkDownstream tasksBlood pressureBlood pressureAF / rhythmAF / rhythm

Datasets linked in the hub; pretraining vs evaluation roles follow the paper.

Architecture
Task models from the benchmark: ResNet18 on image representations, TCN (+MLP) on raw time series, MLP on features (checkpoint file names); full model zoo in paper
Inputs
PPG
Approach
code release · open weights · benchmark · cuffless BP · AF · no licence

Datasets· 2

Cite

APA

QUMPHY D1 PPG benchmark checkpoints (BP + AF; ResNet18 image / TCN / MLP). (2025).

BibTeX

@misc{qumphyd1ppgbench2025qumphy,
  title         = {{QUMPHY D1 PPG benchmark checkpoints (BP + AF; ResNet18 image / TCN / MLP)}},
  year          = {2025}
}

Built from the hub record. Check it against the publisher before submitting.

History

  1. Record updated

  2. Added to the ChatPPG research hub

Connections

Open in the graph explorer
Dataset: PulseDB v2.0 (uses dataset)PulseDB v2.0PulseDB v2…Dataset: Stanford wrist PPG AF (DeepBeat) (uses dataset)Stanford wrist PPG…Stanford w…Signal: PPG (uses signal)PPGPPGQUMPHY D1 PPG benchmark check…QUMPHY D1 PPG ben…
Signals 1
PPG

Same tasks, data and authors

Papers

Models

Datasets