An intelligent IoT-Enabled healthcare system for diabetes prediction using machine learning algorithms

Machine learning (ML) and Internet of Things (IoT) technologies have contributed significantly to various disciplines, i.e., healthcare applications. IoT sensors measure vital data from diabetic patients, while ML can be used to analyze and predict diabetic patients. This paper proposes an innovative intelligent healthcare system (IHS) to predict diabetic patients. This system utilizes the IoT technology and four ML algorithms: random forest (RF), decision tree (DT), k-nearest neighbor (KNN), and artificial neural network (ANN). These algorithms can help in detecting the presence of chronic diseases in patients early and accurately. The algorithms are trained on a diabetic dataset with 2,000 samples. This dataset includes eight features: glucose, pregnancies, blood pressure, insulin, skin thickness, body mass index (BMI), diabetes pedigree function, and age. The proposed algorithms use a Scikit-learn framework with a hyperparameter tuning approach. Experimentally, the performance is evaluated using receiver operating characteristic curves (ROCCs) and error matrices. The results show that the ANN algorithm performs better than RF, DT, and KNN in predicting diabetic patients. The ANN algorithm provides the highest performance, with a testing accuracy (TA) of 99.25%, followed by the RF with a TA of 98.25%, then the DT with a TA of 97.80%, while the KNN demonstrates the worst TA of 96.20%. Machine learning algorithms enable doctors to detect diabetes risk earlier and more accurately. Study compares different methods to find which predicts diabetes best. Results support creating a user-friendly smart application for real-time diabetes monitoring.