Paperdoi:10.5281/zenodo.24184818
Bounded residual learning for monitor-reported stroke volume from ECG and photoplethysmography: analysis code and aggregate outputs
Zenodo code/results release; manuscript stated as Physiological Measurement (status not public)
TL;DR
Frozen-protocol code and outputs for estimating monitor-reported stroke volume from ECG+PPG features on VitalDB (EV1000, 543 subjects). Subject-equal MAE is 13.6 mL for ridge vs 18.7 mL for a training-median baseline, and the physiology-bounded residual model is no better (14.1 mL). Transfer to Vigileo with no update: 12.8–12.9 mL vs 15.4 mL baseline.
- Form factor: perioperative monitor (not wearable)
Abstract
Analysis code, frozen study configuration (protocol v1.3) and aggregate outputs for"Bounded residual learning for monitor-reported stroke volume from ECG andphotoplethysmography".
Contents:- src/ — analysis code (windowing, feature extraction, physiological baseline, bounded residual model, comparators, subject-equal metrics, transport).- protocol/ — frozen study configuration (protocol v1.3; frozen_on 2026-09-01) and run provenance.- results/nested/ — EV1000 subject-grouped nested validation: per-window out-of-fold predictions and metric summaries for the ECG-PPG primary, paired ECG-PPG and paired ART analyses.- results/transport/ — locked zero-update Vigileo evaluation, recalibration audit and predictions.- results/ablations/v1_3/ — A1 (no BMI), A2 (no decay) and A4 (window-equal weights) ablation runs.- results/innovation/v1_3/ — comparator set including the demographics-and- heart-rate ridge model and the negative control (cross-subject waveform feature permutation).- results/sensitivity_5min/ — 300 s sampling-interval sensitivity analysis.- results/windows/ — stage-stratified exclusion logs and the stage-01 cohort status snapshot.
Environment: Python 3.12 with pinned dependencies in requirements-lock.txt.VitalDB source data are not redistributed and are available through the officialaccess process at https://vitaldb.net.
Funding: Anhui Provincial Department of Education Scientific Research Project(Natural Science, Key Project), Grant No. 2025AHGXZK31097.
Verbatim from DataCite (opens in a new tab) · Licence: https://opensource.org/licenses/MIT
Results on ChatPPG leaderboards
All leaderboardsNo results from this paper on ChatPPG leaderboards yet. Browse the leaderboards.
Tasks & signals
Tasks
Signals
Keywords
Datasets used
How it works
Cite this paper
Zhou, Y., Wu, Z., & Zhang, Q. (2026). Bounded residual learning for monitor-reported stroke volume from ECG and photoplethysmography: analysis code and aggregate outputs. Zenodo code/results release. https://doi.org/10.5281/zenodo.24184818
Zhou, Yang, Zhenhua Wu, and Qiao Zhang. 2026. "Bounded residual learning for monitor-reported stroke volume from ECG and photoplethysmography: analysis code and aggregate outputs." Zenodo code/results release. https://doi.org/10.5281/zenodo.24184818.
Zhou, Y., Wu, Z. and Zhang, Q. (2026) 'Bounded residual learning for monitor-reported stroke volume from ECG and photoplethysmography: analysis code and aggregate outputs', Zenodo code/results release. doi: 10.5281/zenodo.24184818.
Zhou, Yang, et al. "Bounded residual learning for monitor-reported stroke volume from ECG and photoplethysmography: analysis code and aggregate outputs." Zenodo code/results release, 2026. https://doi.org/10.5281/zenodo.24184818.
Y. Zhou, Z. Wu, and Q. Zhang, "Bounded residual learning for monitor-reported stroke volume from ECG and photoplethysmography: analysis code and aggregate outputs," Zenodo code/results release, 2026, doi: 10.5281/zenodo.24184818.
@article{zhou2026bounded,
title = {{Bounded residual learning for monitor-reported stroke volume from ECG and photoplethysmography: analysis code and aggregate outputs}},
author = {Zhou, Yang and Wu, Zhenhua and Zhang, Qiao},
year = {2026},
journal = {Zenodo code/results release},
doi = {10.5281/zenodo.24184818}
}Added to the ChatPPG research hub