Development of prediction model for trauma assessment using electronic medical records

Abbreviated Injury Score (AIS) and Injury Severity Score (ISS) scores are used to measure the severity of trauma patients in the emergency department, but they have several problems such as a calculation complexity. In this study, we developed a mortality prediction model of trauma patients using the data from electronic medical records and compared it with a model using AIS/ISS scores. This is a prognostic study using the data of patients who were admitted to Hitachi General Hospital Emergency and Critical Care Center from April 2018 to March 2019. The features were age, sex, vital signs, and clinical diagnoses, and the outcome was in-hospital death. Of 337 eligible patients, 11 died during the hospitalization. The predictive performance of our model was comparable to that of the AIS/ISS scores model (AUC 0.912 vs 0.961). Clinical diagnoses were important in predicting the mortality rate. Our study suggests that a trauma severity index calculated by the predicting model using information from electronic medical records might replace AIS/ISS score.