Early Prediction of Pre-Diabetes Using Machine Learning

Abstract

With the advancement of computer and mobile technologies, mobile heath (mHealth) can be leveraged for patient self-management, patient diagnosis, and determining the possibility of being affected by some disease. Diabetes is a chronic and major disease wherein the body’s sugar content is very high over a prolonged period of time and is possibly reaching epidemic proportions in the world. Although there are some mobile applications keeping track of calories, sugar intake, medicine doses, blood glucose, blood pressure, there is no application exceptionally developed to analyze the risk of being diabetic or to identify if a patient is pre-diabetic. Therefore, the objective of this work is to develop an intelligent mHealth application to assess a person's possibility of being pre-diabetic, diabetic & nondiabetic. The proposed approach uses machine learning for prediction and demonstrates promising results.

Keywords: Pre-Diabetes, Machine Learning, mHealth, Random Forest, Support Vector Machine (SVM), Decision Tree