BackgroundThe purpose of this study was to build a model of machine learning (ML) for the prediction of mortality in patients with isolated moderate and severe traumatic brain injury (TBI).MethodsHospitalized adult patients registered in the Trauma Registry System between January 2009 and December 2015 were enrolled in this study. Only patients with an Abbreviated Injury Scale (AIS) score ≥ 3 points related to head injuries were included in this study. A total of 1734 (1564 survival and 170 non-survival) and 325 (293 survival and 32 non-survival) patients were included in the training and test sets, respectively.ResultsUsing demographics and injury characteristics, as well as patient laboratory data, predictive tools (e.g., logistic regression [LR], support vector machine [SVM], decision tree [DT], naive Bayes [NB], and artificial neural networks [ANN]) were used to determine the mortality of individual patients. The predictive performance was evaluated by accuracy, sensitivity, and specificity, as well as by area under the curve (AUC) measures of receiver operator characteristic curves. In the training set, all five ML models had a specificity of more than 90% and all ML models (except the NB) achieved an accuracy of more than 90%. Among them, the ANN had the highest sensitivity (80.59%) in mortality prediction. Regarding performance, the ANN had the highest AUC (0.968), followed by the LR (0.942), SVM (0.935), NB (0.908), and DT (0.872). In the test set, the ANN had the highest sensitivity (84.38%) in mortality prediction, followed by the SVM (65.63%), LR (59.38%), NB (59.38%), and DT (43.75%).ConclusionsThe ANN model provided the best prediction of mortality for patients with isolated moderate and severe TBI.

Objectives: The shock index (SI) and its derivations, the modified shock index (MSI) and the age shock index (Age SI), have been used to identify trauma patients with unstable hemodynamic status. The aim of this study was to evaluate their use in predicting the requirement for massive transfusion (MT) in trauma patients upon arrival at the hospital. Participants: A patient receiving transfusion of 10 or more units of packed red blood cells or whole blood within 24 h of arrival at the emergency department was defined as having received MT. Detailed data of 2490 patients hospitalized for trauma between 1 January 2009, and 31 December 2014, who had received blood transfusion within 24 h of arrival at the emergency department, were retrieved from the Trauma Registry System of a level I regional trauma center. These included 99 patients who received MT and 2391 patients who did not. Patients with incomplete registration data were excluded from the study. The two-sided Fisher exact test or Pearson chi-square test were used to compare categorical data. The unpaired Student t-test was used to analyze normally distributed continuous data, and the Mann-Whitney U-test was used to compare non-normally distributed data. Parameters including systolic blood pressure (SBP), heart rate (HR), hemoglobin level (Hb), base deficit (BD), SI, MSI, and Age SI that could provide cut-off points for predicting the patients’ probability of receiving MT were identified by the development of specific receiver operating characteristic (ROC) curves. High accuracy was defined as an area under the curve (AUC) of more than 0.9, moderate accuracy was defined as an AUC between 0.9 and 0.7, and low accuracy was defined as an AUC less than 0.7. Results: In addition to a significantly higher Injury Severity Score (ISS) and worse outcome, the patients requiring MT presented with a significantly higher HR and lower SBP, Hb, and BD, as well as significantly increased SI, MSI, and Age SI. Among these, only four parameters (SBP, BD, SI, and MSI) had a discriminating power of moderate accuracy (AUC > 0.7) as would be expected. A SI of 0.95 and a MSI of 1.15 were identified as the cut-off points for predicting the requirement of MT, with an AUC of 0.760 (sensitivity: 0.563 and specificity: 0.876) and 0.756 (sensitivity: 0.615 and specificity: 0.823), respectively. However, in the groups of patients with comorbidities such as hypertension, diabetes mellitus, or coronary artery disease, the discriminating power of these three indices in predicting the requirement of MT was compromised. Conclusions: This study reveals that the SI is moderately accurate in predicting the need for MT. However, this predictive power may be compromised in patients with HTN, DM or CAD. Moreover, the more complex calculations of MSI and Age SI failed to provide better discriminating power than the SI.

Jejunal flap has a significantly lower rate of stricture for reconstruction of circumferential pharyngeal defects when compared with RFF or ALT flaps.