基于云端计算的急性脑梗死后出血性转化的预测模型研究  

Study on the prediction model of hemorrhagic transformation after acutecerebral infarction based on cloud computing

作　　者：曹茂盛 陈瑶瑶 葛胜[1] 薛晓丽 葛颖超 CAO Maosheng;CHEN Yaoyao;GE Sheng;XUE Xiaoli;GE Yingchao(Affiliated Qidong Hospital of Nantong University/Qidong People′s Hospital/Qidong Liver Cancer Institute,Nantong,Jiangsu 226200,China)

机构地区：[1]南通大学附属启东医院/启东市人民医院/启东肝癌防治研究所,江苏南通226200

出　　处：《现代医药卫生》2024年第21期3637-3642,共6页Journal of Modern Medicine & Health

基　　金：江苏省南通市卫生健康委员会青年课题(QNZ2022084);南通大学临床医学专项项目(2022LY017)。

摘　　要：目的构建基于云端服务器的急性脑梗死后出血性转化(HT)的预测模型。方法收集2020年1月至2023年5月该院住院的620例急性脑梗死患者的临床资料,通过随机分组方式分为开发队列(433例)和验证队列(187例)。同时,根据是否发生HT,将开发队列患者分为HT组(64例)和非HT组(369例)。对不同队列及组别天冬氨酸氨基转移酶/丙氨酸氨基转移酶比值(AAR)、血小板计数(PLT)、中性粒细胞/淋巴细胞比值(NLR)、纤维蛋白原(Fib)、总胆固醇(TC)、甘油三酯(TG)、高密度脂蛋白(HDL)、低密度脂蛋白(LDL)等临床资料进行比较,分析HT的影响因素,构建急性脑梗死后HT的预测模型并进行验证。结果开发队列与验证队列在血糖、清蛋白、AAR、Fib、HDL、LDL、镁水平及梗死面积方面比较,差异有统计学意义(P<0.1)。HT组和非HT组在年龄、糖尿病史、房颤史、发病时间、NIHSS评分、血糖、AAR、PLT、TC、TG、HDL、LDL、梗死部位、梗死面积、静脉溶栓方面比较,差异有统计学意义(P<0.1)。Lasso回归分析结果显示,房颤史、梗死面积、NIHSS评分及血糖、HDL水平为发生HT的危险性因素(P<0.1),而LDL、PLT水平为发生HT的保护性因素(P<0.1)。内部验证结果显示,模型bootstrap的曲线下面积(AUC)为0.947(95%CI:0.925~0.969),校准良好(Hosmer-Lemeshow检验的P=0.580,Brier评分为0.069)。外部验证结果显示,AUC为0.916(95%CI:0.857~0.975),校准良好(Hosmer-Lemeshow检验的P=0.596,Brier评分为0.076)。结论基于云端服务器的预测模型可以快速、准确地预测急性脑梗死患者发生HT的风险,有助于进行更准确的个体治疗。Objective To construct a prediction model of hemorrhagic transformation(HT)after acute cerebral infarction based on cloud server.Methods The clinical data of 620 patients with acute cerebral infarction admitted to our hospital from January 2020 to May 2023 were collected and randomly divided into development cohort(433 cases)and validation cohort(187 cases).At the same time,the patients were divided into HT group(64 cases)and non-HT group(369 cases)according to whether HT occurred.The clinical data of aspartate aminotransferase/alanine aminotransferase ratio(AAR),platelet count(PLT),neutrophil/lymphocyte ratio(NLR),fibrinogen(Fib),total cholesterol(TC),triglyceride(TG),high density lipoprotein(HDL)and low density lipoprotein(LDL)in different cohorts and groups were compared.The influencing factors of HT were analyzed,and the prediction model of HT after acute cerebral infarction was constructed and verified.Results There were significant differences in blood glucose,albumin,AAR,Fib,HDL,LDL,magnesium levels and infarct size between the development cohort and the validation cohort(P<0.1).There were significant differences in age,history of diabetes,history of atrial fibrillation,onset time,NIHSS score,blood glucose,AAR,PLT,TC,TG,HDL,LDL,infarct location,infarct size and intravenous thrombolysis between HT group and non-HT group(P<0.1).The results of Lasso regression analysis showed that the history of atrial fibrillation,infarct size,NIHSS score,blood glucose and HDL levels were risk factors for HT(P<0.1),while LDL and PLT levels were protective factors for HT(P<0.1).The results of internal validation showed that the area under the curve(AUC)of the model bootstrap was 0.947(95%CI:0.925-0.969),and the calibration was good(Hosmer-Lemeshow test P value was 0.580,Brier score was 0.069).The results of external validation showed that the AUC was 0.916(95%CI:0.857-0.975),and the calibration was good(Hosmer-Lemeshow test P value was 0.596,Brier score was 0.076).Conclusion The prediction model based on cloud server can quic

关 键 词：出血性转化 急性脑梗死 预测模型 回顾性研究 

分 类 号：R743.33[医药卫生—神经病学与精神病学]