Data visualization and predictive modeling for identifying comorbidities in diabetic patients

Diabetes is one of the most common chronic diseases in the world. Diabetic patients are also more susceptible to develop additional comorbidities over time even causing death. This makes it essential to identify the risk of developing comorbidities as early as possible for effective diabetes management and to reduce the burden on healthcare system. Large volumes of clinical data which has been collected over the years has potential to be translated into meaningful information to enable healthcare professionals gain insights into diabetic patient comorbidities. This research has two key contributions. First, an interactive diabetes dashboard is developed in which the data is integrated and shown in the form of visually appealing charts, graphs and tables. The dashboard displays aggregated results with drilldown capabilities to allow navigation at finer granularities of various metrics. Second, predictive models are built to forecast the likelihood of one of the three common comorbidities for diabetic patients – Benign Hypertension, Congestive Heart Failure, and Acute Renal Failure. The models use advanced data mining algorithms such as Logistic Regression, Neural Network, CHAID, Bayesian Network, Random Forest and Ensemble. Results from these models are also incorporated into an interactive assessment tool that has the ability to take user input and predict the likelihood of one of these comorbidities. Northern Health (NH) dataset consisting exclusively of diabetic patients is used for this research.