Estimation of Heart Rate Using Regression Models and Artificial Neural Network in Middle-Aged Adults

Area of Science:

• Cardiology
• Biomedical Engineering
• Exercise Physiology

Background:

• Heart rate is a key clinical indicator for cardiovascular assessment.
• Existing maximal heart rate prediction models primarily rely on age, often overlooking other influential factors.
• Accurate maximal heart rate estimation is crucial for personalized exercise prescription and clinical evaluations.

Purpose of the Study:

• To develop and evaluate advanced models for maximal heart rate estimation.
• To investigate the predictive power of multiple physiological and demographic factors beyond age.
• To compare the accuracy of multivariate regression and artificial neural networks (ANN) against traditional age-based methods.

Main Methods:

• A cohort of 121 middle-aged adults (average age 57.2 years) underwent maximal exercise testing on a power bike.
• Physiological data including ambulatory blood pressure, electrocardiography, and gas metabolic analysis were continuously monitored.
• Multivariate regression and artificial neural network (ANN) models were employed, incorporating six participant characteristics for analysis.

Main Results:

• The multivariate regression model achieved an estimation accuracy of 9.74%, while the ANN model reached 9.42%.
• Both advanced models demonstrated superior accuracy compared to the traditional age-based model, which had an accuracy of 10.31%.
• Key predictors identified for enhanced accuracy included age, resting heart rate (RHR), and second-order heart rate (SOHR).

Conclusions:

• Multivariate regression and ANN models offer more precise maximal heart rate estimations than conventional age-only formulas.
• Incorporating resting heart rate and second-order heart rate significantly improves prediction accuracy.
• These findings support the use of comprehensive data-driven approaches for personalized cardiovascular assessments.