Physiological-model-based neural network for modeling the metabolic-heart rate relationship during physical activities

Yaowen Zhang; Libera Fresiello; P. H. Veltink; Dirk Donker; Y. Wang

doi:10.1016/j.cmpb.2026.109240

January 1, 2026Computer Methods and Programs in Biomedicine

Physiological-model-based neural network for modeling the metabolic-heart rate relationship during physical activities

View Full Paper

Ask AI

Bookmark

Share

Key Points

Develop and validate a physiological-model-based neural network (PMB-NN) to capture the metabolic-heart rate relationship during physical activities for personalized cardiac tracking.
Trained and tested the framework on individual datasets from 25 participants performing resting, cycling, and running protocols.
Embedded biological constraints from a simplified metabolic-heart rate model into the neural network and compared performance against a benchmark fully connected neural network (FCNN) and a physiological model (PM).
PMB-NN achieved a median R2 score of 0.88, an RMSE of 9.96 bpm, and an MAE of 8.87 bpm, performing on par with FCNN while significantly outperforming PM alone (p < 0.001).
Model plausibility for heart rate and oxygen uptake coupling reached rho = 1, significantly exceeding both FCNN (p = 0.028) and PM (p < 0.001).
Sensitivity analysis demonstrated model robustness, maintaining an RMSE within 15 bpm despite up to 20% Gaussian noise, 4% outliers, and an 18 s time lag.

Structured PICO

P

Population

25 participants (aged 12 to 40, median 18) from two public datasets (WEEE and ACTES) engaged in physical activities including resting, cycling, and running.

I

Intervention

Physiological-model-based neural network (PMB-NN) framework to model the oxygen uptake (VO2)-heart rate (HR) relationship

C

Comparator

Fully connected neural network (FCNN) and a simplified physiological model (PM)

O

Outcome

Heart rate estimation performance evaluated across numerical accuracy, physiological plausibility, and physiological interpretability

Implication

The PMB-NN framework accurately estimates individualized heart rate from oxygen uptake during physical activities, providing a foundation for personalized digital twins in cardiac monitoring.

Abstract

This study confirms the validity of the PMB-NN framework using precise metabolic inputs. This foundational validation enables future integration with wearable-based V̇O₂ estimation systems, ultimately paving the way for personalized, real-time cardiac monitoring during daily life physical activities to enhance HF risk detection.

Discussion

No takes yet. Share an insight, caveat, or question.

Cite This Study

Zhang et al. (2026) studied this question.

synapsesocial.com/papers/6963220591e05aa366cb866c https://doi.org/10.1016/j.cmpb.2026.109240

Topics

Artificial intelligence in cardiology

View Full Paper

Ask AI

Bookmark

Share

Physiological-model-based neural network for modeling the metabolic-heart rate relationship during physical activities

Key Points

Structured PICO

Implication

Abstract

Discussion

Cite This Study

Topics

Physiological-model-based neural network for modeling the metabolic-heart rate relationship during physical activities

Key Points

Structured PICO

Implication

Abstract

Discussion

Cite This Study

Keep reading

Topics

Keep reading