To date, there are no reviews on machine learning (ML) for predicting outcomes in trauma. Consequently, it remains unclear as to how ML-based prediction models compare in the triage and assessment of trauma patients. The objective of this review was to survey and identify studies involving ML for predicting outcomes in trauma, with the hypothesis that models predicting similar outcomes may share common features but the performance of ML in these studies will differ greatly. MEDLINE and other databases were searched for studies involving trauma and ML. Sixty-five observational studies involving ML for the prediction of trauma outcomes met inclusion criteria. In total 2,433,180 patients were included in the studies. The studies focused on prediction of the following outcome measures: survival/mortality (n = 34), morbidity/shock/hemorrhage (n = 12), hospital length of stay (n = 7), hospital admission/triage (n = 6), traumatic brain injury (n = 4), life-saving interventions (n = 5), post-traumatic stress disorder (n = 4), and transfusion (n = 1). Six studies were prospective observational studies. Of the 65 studies, 33 used artificial neural networks for prediction. Importantly, most studies demonstrated the benefits of ML models. However, algorithm performance was assessed differently by different authors. Sensitivity-specificity gap values varied greatly from 0.035 to 0.927. Notably, studies shared many features for model development. A common ML feature base may be determined for predicting outcomes in trauma. However, the impact of ML will require further validation in prospective observational studies and randomized clinical trials, establishment of common performance criteria, and high-quality evidence about clinical and economic impacts before ML can be widely accepted in practice.

To date, no studies have attempted to utilize data from a combination of vital signs, heart rate variability and complexity (HRV, HRC), as well as machine learning (ML), for identifying the need for lifesaving interventions (LSIs) in trauma patients. The objectives of this study were to examine the utility of the above for identifying LSI needs and compare different LSI-associated models, with the hypothesis that an ML model would be superior in performance over multivariate logistic regression models. One hundred four patients transported from the injury scene via helicopter were selected for the study. A wireless vital signs monitor was attached to the patient's arm and used to capture physiologic data, including HRV and HRC. The power of vital sign measurements, HRV, HRC, and Glasgow Coma Scale score (GCS) to identify patients requiring LSIs was estimated using multivariate logistic regression and ML. Receiver operating characteristic (ROC) curves were also obtained. Thirty-two patients underwent 75 LSIs. After logistic regression, ROC curves demonstrated better identification for LSIs using heart rate (HR) and HRC (area under the curve [AUC] of 0.81) than using HR alone (AUC of 0.73). Likewise, ROC curves demonstrated better identification for LSIs using GCS and HRC (AUC of 0.94) than using GCS and HR (AUC of 0.92). Importantly, ROC curves demonstrated that an ML model using HR, GCS, and HRC (AUC of 0.99) had superior performance over multivariate logistic regression models for identifying the need for LSIs in trauma patients. Development of computer decision support systems should utilize vital signs, HRC, and ML in order to achieve more accurate diagnostic capabilities, such as identification of needs for LSIs in trauma patients.

Epidemiologic/prognostic study, level III.