Forecasting Diabetes: Leveraging ML for Early Detection and Prevention

Diabetes is one of the increasing chronic diseases that must be treated well as soon as it is detected. In today's world, diabetes is the most occurring disease. Diabetes is mainly of two types: Type 1 and Type 2 diabetes. The majority of the population is affected by Type 2 diabetes. Over the period, Type 2 Diabetes causes severe damage. Similarly, Type 1 can also cause serious damage, but its prevention is not known yet. So, modern technologies like machine learning would be helpful in predicting diabetes status. In the conventional healthcare system, the diagnosis of this disease primarily depends upon the test reports and the doctor's experience. It may not be accurate every time. Machine learning is a method of classifying different instances by utilizing the knowledge gained using historical data. In this work, the diabetes patient dataset available in the Kaggle repository is used. Different preprocessing methods are used for data cleansing efficiently. This cleaned data can be used for model training. After applying all classification algorithms, the three machine learning algorithms models, such as Logistic Regression (LR), Random Forest (RF), and KNN, are combined to form a hybrid classifier using stacking-based ensemble learning. The proposed system has shown the classification accuracy of 97.57%.