Artificial Intelligence enabled Web-Based Prediction of Diabetes using Machine Learning Approach

Healthcare is the preservation of health that includes diagnosis, surgery, therapy, cure and other related to health of the people. With the advancement of technology, healthcare is enhanced by practicing smart healthcare, e-health and m-Health. In recent years, computerized physician consultant has become popular for the improvement of people's health and this requires a wide range of study of disease that emerges to the clinical decision support system. In this work, an early stage diabetes risk prediction dataset is trained with supervised machine learning and categorized with unsupervised machine learning. The diabetes is classified by best accuracy among supervised machine learning algorithms for a new patient examination. A Web application is created to predict early stage risk of diabetes by classifying results based on the questionnaire of the patient without testing kit using machine learning. Also, the result is analyzed and predicted the chance of having positive or negative diabetes based on the examination falls into clusters by unsupervised machine learning. The evaluation of the prediction is observed to improve the accuracy further with a deep learning approach.