Rationalization of diagnosis and prediction of traumatic brain injury outcomes by blood biomarkers

The purpose of the research is rationalization of diagnosis and prediction of traumatic brain injury outcomes by blood biomarkers. Methods. In 125 examined mature (45–59 years old) and elderly (60–74 years old) patients with mild and moderate TBI, blood counts were studied on the 12th day after receiving it. A general blood test was performed using an automatic analyzer on a GS480A device (China), and a biochemical blood test was performed on a THERMO FISHER SCIENTIFIC Konelab Prime 30 device (Pharmа, Russia). Using a one-factor regression analysis of the studied 13 blood parameters, diagnostic and prognostic significance for 11 variables was revealed. To assess the quality of the predictive multivariate regression model, ROC analysis (Receiver Operator Characteristic) was used, and the area under the curve (AUC) was used to assess the discrimination of the model. Results . In the multifactorial regression analysis in the uncorrected model, all 11 variables with the highest beta coefficient for blood levels of potassium, leukocytes, glucose, lymphocytes, and glucose-to-potassium ratio retained diagnostic and prognostic significance. At the same time, the gender- and age–adjusted multifactorial regression model included only 7 variables, and taking into account the most significant ones, a predictive model was developed: y = 7,561 + 2,652x 1 – 2,848x 2 + 2,458x 3 + 2,573x 4 . The prognostic value of the created model showed that the AUC is 0,725 (p = 0,0012) with a sensitivity of 62,875% and a specificity of 71,896%. Conclusion . The created model is of sufficient quality and can be used to diagnose and predict adverse outcomes of traumatic brain injury.