Predicting Diabetes Mellitus and its Complications through a Graph-Based Risk Scoring System

Abstract

It is vital to estimate and predict the chronological risk rate of individuals of diabetes mellitus and its complications through non-invasive or minimally invasive methods. Data mining and machine learning techniques are applied to health data repositories to achieve this goal. Although past studies have used various combinations of technologies for the assessment and prediction of diabetes and its complications, there is a lack of attention to combining temporal data with a visual representation assessment technique, which can be widely accepted. Further, prediction of risk throughout the lifetime of an individual in a chronological manner by considering their future changes with respect to the characteristics of a similar cohort is something worth contemplating for accurate risk prediction models. We aim to introduce a simple, powerful visualization technique to self-monitoring, which will be highly beneficial in enhancing the health care management sector through empowering self-care management and policymaking. The system will effectively impact the progression of diabetes and its complications by early forecasting the risk without the aid of professional physician knowledge which would help to reduce the burden of the disease while saving the expenditures of diabetes mellitus.