Predicting Mortality in Diabetic ICU Patients Using Machine Learning and Severity Indices Anand, Rajsavi creator Stey, Paul creator Jain, Sukrit creator Biron, Dustin R. creator Bhatt, Harikrishna creator Monteiro, Kristina creator Feller, Edward creator Ranney, Megan L. creator Sarkar, Indra Neil creator Chen, Elizabeth S. creator text posters 2017 eng Biomedical Informatics All rights reserved Diabetes constitutes a significant health problem that leads directly to many long term health problems including renal, cardiovascular, and neuropathic complications that can result in increased health care costs, as well as risk of ICU stay and mortality. Using the MIMIC III database, a Beth Israel ICU database from 2002-2012, machine learning and binomial logistic regression modeling were applied to test numerous predictive algorithms to predict risk of mortality. The final models achieved good fit with strong AUC values of 0.787 and 0.785 respectively. Additionally, this study demonstrated that robust classification can be done as a combination of five variables to predict risk as compared with many other machine learning models that require nearly 35 variables for similar risk assessment and prediction Predictive modelling