Instantaneous Heart Rate-based Automated Monitoring of Hypertension using Machine Learning

Abstract

Hypertension is a serious underlying health condition that can cause a number of severe diseases (such as sudden cardiac events, strokes, etc.) if left untreated or not detected. Wireless watch-style health trackers continuously monitor physiological data and activity that can assist in developing predictive and diagnosis systems to inform vulnerable individuals in real-time. Artificial Intelligence techniques can be extremely beneficial in learning from the existing medical data of hypertensive patients and creating implicit relationships between various relevant physiological parameters for preliminary early diagnosis of hypertension. This study investigates and compares the efficacy of various machine learning techniques for detecting hypertension condition using Instantaneous Heart Rates (IHR). The CNN-LSTM and Bi-LSTM architecture demonstrated better performance for classifying hypertension based on IHR values. The models developed in this study can be incorporated into mHealth or telemedicine applications to detect hypertension and alert the users.