Abstract
Statistical theory indicates that a flexible model can attain a lower generalization error than an inflexible model, provided that the setting is appropriate. This is highly relevant in the context of mortality risk prediction for trauma patients, as researchers have focused almost exclusively on the use of linear models for risk prediction, and linear models may be too inflexible to capture the potentially complex relationships in trauma data. Due to this, we propose an ensemble machine learning model, the Trauma Severity Model (TSM), for risk prediction. In order to empirically validate TSM's predictive performance, this study compares TSM to three established risk prediction models: the Bayesian Logistic Injury Severity Score (BLISS), the Harborview Assessment for Risk of Mortality (HARM), and the Trauma Mortality Prediction Model (TMPM-ICD9). Our results indicate that TSM has superior predictive performance, and thereby provides improved risk prediction.