| 摘要: |
| 目的 基于形态学、血流动力学参数与临床因素构建开颅夹闭术中颅内动脉瘤（Intracranial aneurysm，IA）破裂风险的预测模型，并验证其预测效能。方法 选取邢台市中心医院2020年1月至2025年4月收治的330例IA患者作为研究对象，按照7:3比例分为训练集与验证集，训练集与验证集均根据开颅夹闭术中是否发生IA破裂分为破裂组与未破裂组，术前行计算机断层扫描血管造影（CTA）术，获取动脉瘤形态学参数[动脉瘤最大径、瘤体长径、瘤体宽径、瘤颈宽度、载瘤动脉直径、瘤体最长径与瘤颈宽度的比值（AR）、瘤体最大长径与载瘤动脉直径的比值（SR）]、血流动力学参数[平均壁面剪切力梯度（WSSGA）、震荡剪切指数（OSI）、标准化表面最大剪切力（NWSSM）]，收集临床资料，构建Logistic回归方程（LR）模型和极端梯度提升（XGBoost）模型，通过受试者工作特征曲线（ROC）、校正曲线评估两种模型的预测效能和预测准确性。结果 训练集231例IA患者中开颅夹闭术中IA破裂发生率为38.10%（88/231）；训练集与验证集中破裂组高血压、颅内动脉粥样硬化占比及系统炎症反应指数（SIRI）、动脉瘤最大径、AR、SR、WSSGA、OSI均高于未破裂组，NWSSM低于未破裂组（P＜0.05）；OSI、SIRI、SR、NWSSM、AR、颅内动脉粥样硬化是开颅夹闭术中IA破裂的独立影响因素（P＜0.05）；ROC曲线显示，XGBoost模型在训练集、验证集中的曲线下方面积（AUC）为0.915（95%CI：0.875～0.955）、0.911（95%CI：0.867～0.956），分别高于LR模型的AUC 0.838（95%CI：0.780～0.897）、0.843（95%CI：0.784～0.901）；校正曲线显示，XGBoost模型在训练集、验证集中的预测准确性较LR模型更高。结论 XGBoost模型预测性能、校准度较高，其识别的影响开颅夹闭术中IA破裂的重要特征依次为OSI、SIRI、SR、NWSSM、AR、颅内动脉粥样硬化，据此能精准识别潜在的高风险个体，为临床手术策略、应急预案决策提供参考依据。 |
| 关键词: 颅内动脉瘤 开颅夹闭术 动脉瘤破裂 风险预测模型 影响因素 |
| DOI： |
| 分类号: |
| 基金项目:邢台市重点研发计划自筹项目（2024ZC194） |
|
| Construct a predictive model for the risk of intracranial aneurysm rupture during craniotomy clipping, and validating its efficacy based on morphology, hemodynamic parameters and clinical factors |
|
wangshengjun
|
|
Xingtai Central Hospital
|
| Abstract: |
| Objective ? To construct a predictive model for the risk of intracranial aneurysm (IA) rupture during craniotomy clipping based on morphology, hemodynamic parameters and clinical factors, and to verify its predictive efficacy. Methods ?A total of 330 patients with IA admitted to Xingtai Central Hospital from January 2020 to April 2025 were selected as the research objects. They were divided into training set and validation set according to the ratio of 7:3. The training set and validation set were divided into ruptured group and unruptured group according to whether IA rupture occurred during craniotomy clipping. Computer tomography angiography (CTA) was performed before operation. The morphological parameters of aneurysms [maximum diameter of aneurysm, length of aneurysm, width of aneurysm, width of aneurysm neck, diameter of parent artery, ratio of maximum diameter of aneurysm to width of aneurysm neck (AR), ratio of maximum length of aneurysm to diameter of parent artery (SR) ] and hemodynamic parameters [ mean wall shear stress gradient (WSSGA), oscillatory shear index (OSI), normalized surface maximum shear stress (NWSSM)] were obtained. Clinical data were collected to construct Logistic regression equation (LR) model and extreme gradient boosting (XGBoost) model. The predictive efficacy and accuracy of the two models were evaluated by receiver operating characteristic curve (ROC) and calibration curve. Results ?The incidence of IA rupture during craniotomy clipping was 38.10 % (88/231) in 231 IA patients in the training set. The proportion of hypertension, intracranial atherosclerosis, systemic inflammatory response index (SIRI), maximum diameter of aneurysm, AR, SR, WSSGA and OSI in the rupture group of training set and validation set were higher than those in the unruptured group, and NWSSM was lower than that in the unruptured group (P<0.05). OSI, SIRI, SR, NWSSM, AR and intracranial atherosclerosis were independent influencing factors of IA rupture during craniotomy clipping (P<0.05). The ROC curve showed that the area under the curve (AUC) of the XGBoost model in the training set and the validation set was 0.915 (95%CI: 0.875-0.955) and 0.911 (95%CI: 0.867-0.956), respectively, which were higher than the AUC of the LR model 0.838 (95%CI: 0.780-0.897) and 0.843 (95%CI: 0.784-0.901). The calibration curve showed that the XGBoost model had higher prediction accuracy than the LR model in the training set and the validation set. |
| Key words: Intracranial aneurysms Craniotomy clipping Aneurysm rupture Risk prediction model Influencing factors |
