【摘要】 目的 构建基于术中造影征象的急性大血管闭塞型缺血性卒中(AIS-LVO)病因预测模型。方法 回顾性纳入2019年5月至2022年5月在吉林大学中日联谊医院接受动脉取栓治疗的急性颈内动脉供血区缺血性卒中患者，收集其人口学特征、危险因素、NIHSS评分、静脉溶栓情况、术中造影征象(截断征、爪形征、轨道征、锥形征)及卒中病因。采用多变量Logistic回归构建病因预测模型，并通过加强Bootstrap法进行内部验证。一致性统计量和校准图用于评价模型的区分度和校准度。结果 本研究共纳入116例患者。多变量Logistic回归分析显示，血脂异常(OR=3.18,95%CI:1.16～9.01,P=0.03)、心房纤颤(OR=0.13,95%CI:0.04～0.37,P<0.01)及锥形征(OR=6.24,95%CI:1.93～22.77,P<0.01)为急性大血管闭塞型缺血性卒中病因的独立预测因素。这些变量被纳入预测模型，经内部验证后，模型的一致性统计量为0.86,显示出较高的区分度。校准图表明，该模型具有良好的校准度。结论 由血脂异常、心房纤颤和锥形征组成的预测模型，可以预测AIS-LVO患者的病因。更多还原

【Abstract】 Objective Develop a predictive model for the etiology of acute ischemic stroke caused by large vessel occlusion(AIS-LVO) based on intraprocedural angiographic signs(IPAS).Methods Consecutive AIS-LVO patients who underwent EVT at China-Japan Union Hospital of Jilin University between May 2019 and May 2022 were included in this study retrospectively.Baseline clinical data, details of IPASs, and outcomes were collected from medical records and DSA images.The prediction model for stroke etiology was developed using multivariate logistic regression analysis.The Concordance statistics(C-statistics) and calibration plot with bootstraps of 500 resamples were used to assess discrimination and calibration of the model.Results 116 patients were included in our study.Dyslipidemia(OR=3.18,95% CI:1.16～9.01,P=0.03),atrial fibrillation(OR=0.13,95% CI:0.04～0.37,P<0.01)and tapered sign(OR=6.24,95% CI:1.93～22.77,P<0.01) were independent predictors of etiology and were integrated into prediction model.The C-statistics was 0.86.Calibration plot revealed that the model exhibited good calibration performance.Conclusion The prediction model integrating dyslipidemia, atrial fibrillation, and the tapered sign demonstrated the ability to predict the etiology of AIS-LVO.更多还原