Abstract—Sepsis is a syndrome that occurs with physiological and biochemical abnormalities induced by severe infection and carries a high mortality and morbidity, therefore the severity of its condition must be interpreted quickly. This study presents the development of a model for the one-year mortality prediction of the patients that are admitted in a ICU with a sepsis diagnosis. 5650 patients extracted from the MIMIC III database (divided in 70% for training and 30% for validation) were evaluated and predictors available from the ICU admission was used to develop a mortality prognosis prediction model based on Bayesian Additive Regression Trees (BART) methodology. Variable importance is also presented. In order to evaluate the predictive power of the model, we used the 1695 admissions of the validation subset, and obtained an area under the Receiver Operating Characteristic curve (AUROC) of 0.7354 (95% Confidence Interval (CI): [0.7118-0.7589]). The presented model outperform the results obtained with Sequential Organ Failure Assessment (SOFA), Oxford Acute Severity of Illness Score (OASIS) and Simplified Acute Physiology Score II (SAPSII) indicators on the same validation subset. Our approach demonstrates the importance of comorbidities for the long-term mortality in patients with sepsis in the ICU and shows that it is possible to obtain a model with adequate predictive capacity from the moment of the admission of a patient.
 
Index Terms—bayesian additive regression trees, prognosis prediction, sepsis, intensive care unit