RPHM: Real-Time Personalized Health Monitoring Using Wearable IoT and AI

Abstract

When the latest technology is integrated with real-time personalized health monitoring, there will be noteworthy improvement in the proactive healthcare. In this paper, a novel and comprehensive framework called Real-time Personalized Health Monitoring (RPHM), Real-time Personalized Health Monitoring is proposed. This framework integrates the wearable IoT sensors, cloud for storing data, Artificial Intelligence (AI) algorithms for analysis, and user feedback to qualify the lifestyle of an individual and to indicate the health status of an individual and recommend to do some micro-actions. The architecture of the proposed system is scalable and modular. It consists of six layers: sensing, transmission, storage, processing, application and feedback learning layers. Two novel concepts are introduced in the proposed system: Personalised Health Index (PHI) and Micro-Action Recommendation Engine. PHI is the weighted score which indicates the holistic wellness which integrates various health parameters, heart rate (25%), glucose (30%), activity (20%), temperature (15%), and calories (10%) for the assessment of health in real-time. Micro-Action Recommendation Engine suggests simple and easy actions based on the health condition of an individual in order to improve the health. Micro-Actions are dependent on the PHI score. Support Vector Machines (SVM), Random Forest, Decision Tree classifiers and Long Short-Term Memory (LSTM) networks are ensembled in the proposed framework for time-series forecasting. The performance of the proposed framework RPHM is compared with AI4FoodDB, Diet Engine, and NUTRIVISION and proved to be better in terms of 14 different performance metrics like 4 classification metrics, 3 forecasting metrics, 3 system latency metrics, 2 system reliability metrics and 2 innovation metrics. The Proposed framework, RPHM achieved 90% + accuracy (Classification Metric), 4.82 Mean Absolute Error (MAE) forecasting (Forecasting Metric), 3.4 s end-to-end delay (System Latency Metric), along with 99.2% system uptime and 0.85 PHI correlation. The incorporation of guidance on customised nutrition along with the physical actions in the RPHM is a key development in digital health technology.