Leveraging Machine Learning in Mist Computing Telemonitoring System for Diabetes Prediction

Big data analytics with the help of cloud computing is one of the emerging areas for processing and analytics in healthcare system. Mist computing is one of the paradigms where edge devices assist the fog node to help reduce latency and increase throughput for assisting at the edge of the client. This paper discusses the emergence of mist computing for mining analytics in big data from medical health applications. The present paper proposed and developed mist computing-based framework, i.e., MistLearn for application of K-means clustering on real-world feature data for detecting diabetes in-home monitoring of patients suffering from diabetes mellitus. We built a prototype using Intel Edison and Raspberry Pi; the embedded microprocessor for MistLearn. The proposed architecture has employed machine learning on a deep learning framework for analysis of pathological feature data that can be obtained from smartwatches worn by the patients with diabetes. The results showed that mist computing holds an immense promise for analysis of medical big data especially in telehealth monitoring of patients.