Construction and validation of a risk prediction model for the prognosis of patients with diffuse brain swelling after traumatic brain injury

Ying Xiyi; Liu Peng

doi:10.3969/j.issn.1002-1949.2025.06.004

Abstract：Objective To explore the influencing factors of prognosis in the patients with diffuse brain swelling after traumatic brain injury(TBI),and to construct a risk prediction model.Methods The data of 149 patients with diffuse brain swelling after TBI admitted to Ningbo Hangzhou Bay Hospital from June 2019 to June 2024 were retrospectively collected.According to the survival status of the patients within 14 days,they were divided into the death group(n=78)and the survival group(n=71).The clinical data of the two groups were compared.The possible influencing factors of prognosis in the patients with diffuse brain swelling after TBI were obtained by multivariate Cox regression analysis.A risk prediction model was constructed based on the results of multivariate Cox regression analysis,and the model was visualized with nomogram.The predictive efficacy of the model was internally verified through the receiver operating characteristic(ROC)curve,calibration curve and decision curve.In addition,the data of 63 patients with diffuse brain swelling after TBI admitted to the hospital from July 2024 to March 2025 were collected as the validation set.The ROC curve was plotted by using the data from the validation set to externally validate the predictive efficacy of the model.Results After preliminary comparison of data between the death group and the survival group,continuous variable indicators with differences[Glasgow coma scale(GCS)score at admission,hemoglobin,neutrophil/lymphocyte ratio(NLR),C-reactive protein,albumin,ferritin,and ubiquitin carboxyterminal hydrolase L1(UCH-L1)]were used to obtain the optimal threshold for predicting mortality risk of the patients through ROC curve.Based on the optimal threshold,the above indicators were classified into binary variables and included in all data for multivariate Cox regression analysis.The results showed that subarachnoid hemorrhage,invasive monitoring of intracranial pressure during treatment,hemoglobin＜112.44g/L,NLR≥9.21,C-reactive protein≥11.34 mg/L,ferritin≥279.15 ng/mL and UCH-L1 ≥33.27 ng/mL were independent risk factors for the prognosis of patients with diffuse brain swelling after TBI(P＜0.05).A risk prediction model was constructed based on the results of multivariate Cox regression,and the ROC curve showed that the area under curve(AUC)of the model predicting the mortality of patients with diffuse brain swelling after TBI was 0.982(95％CI 0.969-0.997),indicating good discrimination.The model calibration curve tended towards the ideal curve with good consistency.Decision curve was drawn with a risk threshold within the range of 0.00-1.00.The clinical net benefit rate of the model was always greater than 0,and the maximum net benefit rate was 0.460.The model had good clinical application value.The AUC of the validation set was 0.974(95％CI 0.942-1.000),indicating that the established model had good external predictive performance.Conclusions The combination of subarachnoid hemorrhage,intracranial pressure monitoring during treatment,hemoglobin,NLR,C-reactive protein,ferritin,and UCH-L1 may all be risk factors for the prognosis of patients with diffuse brain swelling after TBI.The risk prediction model constructed based on these indexes may have good predictive efficacy for the short-term prognosis risk of patients.