SMART DIABETES CARE: INTEGRATING IOT WITH DATA ANALYTICS FOR PERSONALIZED HEALTH INSIGHTS

Abstract

Diabetes being a chronic disease is a constant subject for surveillance, detection, and personalized care. With the infusion of IoT technologies in healthcare, there exists an opportunity to develop intelligent systems to collect such real-time physiological data as glucose levels, heart rate, and physical activity via wearable devices such as Continuous Glucose Monitors (CGMs) and fitness trackers. Nevertheless, it presents an enormous challenge for traditional diagnosis systems due to the varied nature of the overwhelming amount of this data. To this effect, this research study proposes a smart diabetes detection framework combining IoT-based data collection and cutting-edge machine learning and deep learning techniques. The system utilizes data cells for pre-processing through Z-Score normalization and KNN imputation to ensure consistency and quality of the data. It employs deep feature extraction by means of Autoencoders to lower dimension and preserve key patterns; these are then flattened into a vector and fed to a Recurrent Neural Network (RNN) classifier, which exploits temporal dependencies varying between features to determine the presence or absence of diabetes accurately. The classification results are analysed and visualized further to forge interpretable insights. Evaluation by performance metrics and user-level analysis demonstrates the efficacy of the system in supporting early diagnosis and personalized management of diabetes. This convergent approach will thus facilitate adaptive, data-driven patient-centric smart healthcare solutions.