Randomly Under Sampled Boosted Tree for Predicting Sepsis From Intensive Care Unit Databases

摘要

Bacterial infection can result in sepsis; a toxic immune response by the body. Although the rate of mortality due to sepsis has fallen within the UK, overall rates remain higher than in Europe. Early detection of sepsis has been linked to elevated successful outcomes. This work focuses on the use of a Random Under Sample (RUS) Boosted Tree for classifying sepsis from intensive care unit databases.Full training set (40,336 subjects) achieved sensitivity and specificity of 53.4% and 83.6% respectively (AUC:77.79%) with 5 fold cross-validation. In the unofficial phase of the challenge the model achieved a normalized utility score of 0.267. The model did not achieve a score on the full test set (team name B-Secur).The results show that the model is capable of detecting sepsis in patients. However, there is more work to be done in order to improve performance. Future work will investigate the use of fixed time windows rather than individual hourly measurements to increase prediction performance.