A novel and intelligent decision-making system for real-time healthcare tracking using commercial wearable data

Abstract

Wearable health devices became popular these days and have become genuinely intertwined with society. Smartwatches and other fitness devices fulfil the consumer needs in continuously tracking human activity, which can further decode to analyse the health parameters like heart rate, blood pressure, blood glucose levels, and many more. Internet of things (IoT) enabled techniques, mobile and desktop-based applications are ameliorating the ease of using these techniques. The applications of the wearables are also transforming the quality of virtual and tele-healthcare to improve, which is a substitute for conventional medical practices. In this paper, we report a descriptive analysis on the progress in modelling the healthcare wearable sensors that impact the imminent healthcare applications in different domains. Also, we made a comparative study on consumer fitness wearable devices to analyse how the device facilitates the ease of usage with other specification comparisons. We recorded data from a consumer wearable fitness device to observe and envisage the user’s effort to accomplish the activity goals each day for maintaining good health. We reported the exploratory analysis of the data obtained from the recordings. Supervised machine learning algorithms are applied to the recorded data and compared the results. Among the supervised algorithms applied, the random forest regression gave us the highest accuracy of 97.88% in predicting the subject’s activity goal for the respective day.