Pioneering Wearable Sensor-Driven Health-Monitoring System for Contagious Disease Prevention With Intelligent Crowd-Counting Models

Abstract

Contagious diseases such as COVID have significantly increased the need for personal health-monitoring systems through wearable devices. This article presents a model based on wearable devices for a health-monitoring system that aims to assist Hajj and Umrah pilgrims with tracking their vitals and providing them with safety advice. Using an advance deep reinforcement learning (DRL) model called deep Q network (DQN) helps to make adaptive decisions on alerts based on the context and historical data records of individual health records and mitigates false alarms. Wearable wristbands equipped with different types of sensors such as temperature, SPO2, time of flight (ToF), and heart rate sensors are employed. The data accumulated from sensors are analyzed periodically for density estimation and safety classification. The classification is based on the level of a threshold. The threshold level is determined based on the distance between persons and the number of persons. The analysis of the sensed data recommends a safe distance (SD) for person-to-person interaction and provides self-assisted health monitoring. The system is evaluated in different intervals between 5 and 60 min. Results reveal that the proposed model effectively improves the data analysis rate by 14.58%, density detection by 16.12%, recommendations by 15.79%, and distortion error by 11.57%.