Effectiveness validation of Physical Exercise Routines for Hypertensive Patients using Wearable Devices and Machine Learning

¹ Universidad de Guayaquil, 090514 Guayaquil, Ecuador
² Universidad de Jaén, 23071 Jaén, Spain
³ Universitat Oberta de Catalunya, 08018 Barcelona, Spain

^* Corresponding author: mariuxi.toapantab@ug.edu.ec

Abstract

High blood pressure is a chronic disorder that consists of increased blood pressure. Research has demonstrated that physical exercise can have effects like the use of some medications in hypertensive patients, leading healthcare professionals to prescribe physical activities for these types of patients commonly. However, hypertensive patients have different physiological realities, so the treatment and prescribed exercises must be personalized and adapted to each one of them. Relevant information has been gathered from bibliographic sources and surveys of medical experts to identify the factors influencing hypertension, commonly used machine learning models in medicine, and clinical datasets of patients with hypertension. This data is used to train predictive models, enabling specialists to assess the effectiveness of prescribed exercises for hypertensive patients. A comparative analysis of machine learning models such as Naive Bayes, Decision Tree, and Logistic Regression was conducted to determine the best model for evaluating the effectiveness of prescribed exercises. Metrics such as accuracy, precision, and recall were used for evaluation. The Decision Tree algorithm achieved the best performance with an accuracy rate of 79%. This evaluated model will be integrated into the platform developed in the subsequent phases of the FCI Research Project.