A Machine Learning Strategy for Internet-of-Things-Enabled Diabetic Prediction to Mitigate Pneumonia Risk

Diabetes has been recently listed as the tenth-largest cause of mortality, accounting for 1.5 million fatalities, and it causes blindness, kidney failure, coronary artery disease, and even diabetic agony. Recent studies have shown that people with type 2 diabetes have an increased mortality risk if they are hospitalized for pneumonia. Non-invasive wearable gadgets and the Internet of Healthcare Things (IoHT) allow for more precise monitoring of blood glucose levels. Intensive care unit patients, expectant mothers, aged and bed-driven people have a constant battle to keep their blood glucose (BG) levels in check, including both hyper and hypoglycemia. Since internet-based healthcare systems heavily rely on technology to ensure adequate glucose control, the subject matters must be addressed cost-effectively. Using Photoplethysmograph and machine learning approaches, this research developed a hybrid non-invasive internet of things-based BG monitoring system to mitigate pneumonia risk. The developed IoT gadget is designed with MAX30102 and MLX90614 for measuring blood glucose and body temperature. The developed hardware, support vector machine (SVM), and k-nearest neighbors (K-NN) algorithm had deployed in Raspberry Pi 4B computer with graphical processing unit (GPU) support. A non-invasive glucose monitor that is more accurate than existing glucose monitoring technology would be a big step forward. This hybrid model offers 81.7% prediction accuracy and adaptability within the clinical laboratory precision criteria range.