Abstract
To accurately forecast heart rate changes during exercise, which is essential for customized health monitoring and improving training regimens, it is necessary to comprehend both the physiological foundations and the technical capacities for data processing. This research utilizes Machine Learning (ML) methodologies to predict heart rate reactions based on physical characteristics and activity variables. Our research focuses on the health and sports aspects of our results, using a comprehensive dataset that includes a wide range of activity types and ambient circumstances across 12,000 sets. We establish a connection between the ability of models such as Linear Regression (LR) and Extreme Gradient Boosting (XGB) to predict outcomes and their practical use in exercise management and optimizing athlete performance. These models accurately forecast variations in heart rate and also provide insights into the cardiovascular demands of various physical activities. Standard metrics measure the effectiveness of these models. The Linear Regression (LR) model achieved a Mean Absolute Error (MAE) of 0.419, a Mean Squared Error (MSE) of 0.294, a Root Mean Squared Error (RMSE) of 0.543, and an R-Squared value of 0.997. On the other hand, the Extreme Gradient Boosting (XGB) Regressor model achieved a Mean Absolute Error (MAE) of 0.421, a Mean Squared Error (MSE) of 0.335, a Root Mean Squared Error (RMSE) of 0.578, and an R-Squared value of 0.996. These metrics demonstrate the usefulness of these models in real-world scenarios. Our study's findings demonstrate that the combination of physiological data and powerful machine learning models may improve an individual's comprehension of fitness levels and the requirements for adaptive training. This study not only adds to the field of computational physiology, but it also aids in the creation of adaptive, real-time therapies for improving health and performance.
Publisher
Asian Research Association