Towards an Internet of Things and Machine Learning-based Health Monitoring and Management System.

Abstract

In urban regions, sedentary life has reached an all-time high due to technological advancements in the last several years. As a consequence, young people are becoming fat and obese. Among the many negative effects of being overweight or obese include diabetes, heart disease, high blood pressure, and a host of other conditions. Recent years have seen a proliferation of machine learning applications in a range of sectors such as financing, banking, social good including forecasting and healthcare. In this work, we specifically touched upon the healthcare domain to develop a system that leverages machine learning and components of internet of things for measuring health levels and thus recommend a proper plan for future health goals of users. A dashboard for tracking progress and customising diet and exercise routines are also part of the system design. Using a range of daily living factors of users, we also include a framework that employs machine learning techniques such as Random Forest, Decision Tree, XGBoost, Extra Trees, and KNN to train models that can predict obesity levels, body weight, and fat percentage levels. In addition, as a part of machine learning subdomain, we analyzed variety of methods to improve the model's accuracy using hyperoptimization algorithms, including gridsearchCV, randomsearchCV, optuna framework and genetic algorithm. The framework is built using Python Flask. Using Internet of Things (IoT)-enabled weighing scales, the framework can also keeps track of the calories and macronutrients that are consumed by the users.