基于血常规和颈动脉斑块构建缺血性脑卒中nomogram风险预测模型

Construction of a nomogram prediction model for ischaemic stroke based on routine blood test and carotid plaque

目的:基于血常规和颈动脉斑块构建一种个性化nomogram风险预测模型预测颈动脉粥样硬化(carotid atherosclerosis, CAS)患者发生缺血性脑卒中(cerebral ischemic stroke, CIS)的风险。方法:选取2021年3月1日至2022年3月1日在上海市第八人民医院神经内科住院的CAS患者214例,收集患者的基本特征、血常规指标及影像学检查数据。根据是否发生缺血性脑卒中分别分为两组,随机抽取全部数据按7∶3的比例拆分为建模组和验证组。采用单因素logistic回归和lasso回归筛选CAS患者发生缺血性脑卒中的独立风险预测因子,将其导入R软件构建nomogram预测模型。ROC曲线下面积(AUC)、校准曲线和DCA决策曲线对模型进行内部验证。结果:单因素logistic回归和lasso回归分析结果显示,红细胞分布宽度、大型血小板比率、血小板计数是CAS患者发生缺血性脑卒中的独立风险预测因子(P<0.05),由于年龄对于CIS具有重要临床意义,最终也将其纳入模型。基于上述预测因子导入R软件构建nomogram预测模型并进行模型内部验证。建模组受试者工作特征曲线下面积(area under the curve, AUC)为0.644,验证组AUC为0.677,表示该nomogram模型预测能力较好。Hosmer-Lemeshow拟合优度检验(P=0.058),表明该模型具有较好的区分度。DCA曲线显示风险阈值为8%~45%时使用该模型具有临床实用价值。结论:本研究构建并验证了一个预测CAS患者发生缺血性脑卒中的nomogram风险预测模型,该模型预测能力和区分能力较好,对临床评估CAS患者发生缺血性脑卒中具有较高的临床实用价值。

Objective:To predict the risk of cerebral ischemic stroke(CIS)in patients with carotid atherosclerosis(CAS)by constructing a personalized nomogram prediction model based on routine blood tests and carotid plaque.Methods:A total of 240 CAS patients who were admitted to the Department of Neurology of Shanghai Eighth People's Hospital from March 1,2021 to March 1,2022 were selected,and the basic characteristics,routine blood indicators and imaging data were collected.The patients were divided into two groups according to whether they had ischemic stroke or not,and all data were randomly selected and split into the modeling and validation group in the ratio of 7:3.The model was internally validated using the area under the ROC curve(AUC),calibration curve and decision curve analysis(DCA).Results:One-way logistic regression and lasso regression analyses showed that red cell distribution width(RDW),platelet-large cell ratio(P-LCR)and platelet count were independent predictors of ischaemic stroke in patients with CAS(P<0.05),and age was eventually included in the model due to its clinical significance for CIS.A nomogram prediction model was constructed based on the above predictors imported into R software and the model was validated internally.The Hosmer-Lemeshow goodness of fit test(P=0.058)indicated that the model had good discrimination.DCA results showed that using the model at risk thresholds of 8%to 45%was of good clinical practice.Conclusion:A nomogram prediction model for ischaemic stroke in CAS patients was constructed and validated in this study,which confirmed that the model had good predictive and discriminatory ability,and it is of high clinical utility in the clinical assessment of ischaemic stroke in CAS patients.