Predictive Analysis of Type 2 Diabetes Using Hybrid ML Model and IoT

Abstract

Diabetes is a severe lifelong disease affecting millions of people around the world. Early detection of diabetes can help doctors to diagnose it in the early stages. With the latest advancements in the field of IoT and sensors, these IoT-enabled sensors can be utilized for the analysis and diagnosis of “type II” diabetes. Using IoT-enabled devices and various sensors, hospitals can now generate data related to patients. The treatment, laboratory examination, and patients’ data can be significant for the identification of “type II” diabetes. These IoT devices can be integrated with machine learning algorithms to make data-driven decisions. This will help in the early detection of diabetes and providing new opportunities for doctors to improve both the patient’s health and experience. In this research work, we are using the Pima Indians Diabetes dataset. We propose a stacking classifier model to identify whether a person has “type II” diabetes or not. Stacking is an ensemble learning technique that uses multiple machine learning algorithms to achieve higher predictive power. The proposed model leverages the power of deep neural networks, decision trees, and ensemble learning. We have used multilayer perceptron (MLP), random forest classifier (RFC), k-nearest neighbors (KNN), and eXtreme Gradient Boosting (XGBoost) as base classifiers and logistic regression as a meta-classifier.