Abstract: Objective To establish an individualized column chart model for predicting the risk of rapid postoperative recovery in patients with traumatic brain injury(TBI). Methods A retrospective analysis was conducted on the clinical data of 240 TBI patients who underwent surgical treatment in our hospital from October 2023 to November 2025.70% of patients were randomly selected as the training group(n=168),and 30% of patients were selected as the validation group(n=72).Patients in the training group were assigned into a rapid recovery group(n=101) and a poor recovery group(n=67) based on whether they recovered quickly after surgery.The logistic regression model was used to screen high-risk factors that affected the rapid postoperative recovery of TBI patients in the training group,and the column chart prediction model was constructed.The predictive performance of the column chart model was evaluated through ROC curve,calibration curve,and Hosmer-Lemeshow goodness of fit test,and external validation was performed using data in validation group. Results There were statistically great differences in age,preoperative GCS score,BIS,length of hospital stay,midline shift,basal ganglia,lesion side,and bleeding site between the rapid recovery group and the poor recovery group(P＜0.05).Logistic regression analysis showed that age>65 years old(OR=1.545,95% CI=1.108~1.849),decreased GCS score(OR=1.392,95% CI=1.256~1.601),decreased BIS(OR=1.919,95% CI=1.869~1.973),and midline shift>15 mm(OR=1.326,95% CI=1.122~1.869) were independent risk factors for poor postoperative recovery in TBI patients(P＜0.05).The area under the ROC curve of the model in the training group was 0.935(95% CI=0.895~0.975),indicating good discriminability;the calibration curve tended towards the ideal curve,and the actual value was in good agreement with the predicted value.The area under the ROC curve of the model in the validation group was 0.929(95% CI=0.873~0.985),indicating good discriminability;the calibration curve tended towards the ideal curve,and the actual value was in good agreement with the predicted value. Conclusion The column chart model constructed based on the four independent risk factors selected by the logistic regression model has good predictive value for the risk of poor postoperative recovery in TBI patients.

Key words: Traumatic brain injury, Rapid postoperative recovery, Individualized prediction, Column chart model

摘要： 目的 建立个体化预测影响颅脑外伤(traumatic brain injury,TBI)患者术后快速康复的风险列线图模型。方法 回顾性分析2023年10月至2025年1月本院收治的240例进行手术治疗的TBI患者的临床资料,随机抽取70%患者为训练组(n=168),30%患者为验证组(n=72),训练组患者根据术后是否快速康复分为快速康复组(n=101)和康复不良组(n=67)。Logistic回归模型用于筛选影响训练组TBI患者术后快速康复的高危因素,以及列线图预测模型的构建,并通过ROC曲线、校准曲线、Hosmer-Lemeshow拟合优度检验评估该列线图模型预测效能,采用验证组数据进行外部验证。结果 快速康复组与康复不良组患者年龄、术前GCS评分、BIS、住院天数、中线移位、基底池、病灶侧、出血部位比较,差异有统计学意义(P＜0.05)。Logistic回归分析结果显示,年龄＞65岁(OR=1.545,95%CI=1.108~1.849)、GCS评分下降(OR=1.392,95%CI=1.256~1.601)、BIS(OR=1.919,95%CI=1.869~1.973)下降、中线移位＞15 mm(OR=1.326,95%CI=1.122~1.869)是TBI患者术后康复不良的独立危险因素(P＜0.05)。模型在训练组中的ROC曲线下面积为0.935(95%CI=0.895~0.975),区分度良好;校正曲线趋近于理想曲线,实际值与预测值一致性良好。模型在验证组中的ROC曲线下面积为0.929(95%CI=0.873~0.985),区分度良好;校正曲线趋近于理想曲线,实际值与预测值一致性良好。结论 基于Logistic回归模型筛选出的4项独立危险因素构建的列线图模型对TBI患者术后康复不良的风险具有较好的预测价值。

关键词: 颅脑外伤, 术后快速康复, 个体化预测, 列线图模型