凝血指标构建脑小血管病总负荷预测模型的价值及验证

Value and validation of coagulation index in prediction model of total load of small cerebral vascular disease

目的基于凝血指标构建脑小血管病(CSVD)总负荷的诺莫图预测模型,分析预测模型的价值并进行验证。方法选取2020-01—2023-12蚌埠医科大学第二附属医院收治的200例CSVD患者,根据影像学评估CSVD总负荷,按总负荷评分(0~4分)分为低负荷组(117例,0~2分)、高负荷组(83例,3~4分)。比较2组一般资料、凝血指标[凝血酶原时间(PT)、纤维蛋白原(FIB)、D-二聚体、血浆凝血酶时间(TT)、活化部分凝血活酶时间(APTT)],采用LASSO回归分析初筛CSVD高负荷的预测因素,通过Logistic分析CSVD高负荷的影响因素,根据影响因素运用R语言构建CSVD高负荷的诺莫图预测模型,并进行模型的评价与验证。结果2组年龄、高血压、新发或既往卒中、糖尿病、高同型半胱氨酸血症(HHcy)情况比较差异有统计学意义(P<0.05)。高负荷组FIB、D-二聚体水平高于低负荷组,PT、APTT小于低负荷组(P<0.05)。CSVD高负荷预测因素初筛选出7个预测变量,Logistic回归分析显示年龄、高血压、新发或既往卒中、FIB、D-二聚体、APTT是CSVD高负荷的独立影响因素(P<0.05)。ROC曲线显示该诺莫图预测的曲线下面积为0.802(95%CI:0.743~0.862),校准曲线显示该诺莫图预测CSVD高负荷风险与实际风险状况基本一致,决策曲线显示该诺莫图具有较好的临床净收益。结论年龄、高血压、新发或既往卒中、FIB、D-二聚体、APTT与CSVD总负荷有关,基于上述凝血指标构建的诺莫图预测模型在CSVD高负荷风险预测中预测效能、临床实用性较高,具有良好区分度、校准度,对临床防治有积极指导意义。

Objective To construct a Nomogram prediction model of total cerebral small vascular disease(CSVD)load based on coagulation index,and analyze the value of the model and verify it.Methods A total of 200 patients with CSVD admitted to the Second Affiliated Hospital of Bengbu Medical University from January 2020 to December 2023 were selected to evaluate the total CSVD load according to imaging,and were divided into low load group(117 cases,0-2 points)and high load group(83 cases,3-4 points)according to total load score(0-4 points).The general data,coagulation indexes(prothrombin time(PT),fibrinogen(FIB),D-dimer,plasma thrombin time(TT),activated partial thomboplastin time(APTT))of the two groups were compared,and the predictive factors for CSVD high load were analyzed by LASSO regression.The influencing factors for CSVD high load were analyzed by Logistic analysis,and the Nomograph prediction model of CSVD high load was constructed by R language according to the influencing factors,and the model was evaluated and verified by receiver operating characteristic(ROC)curve,calibration curve and decision curve analysis(DCA).Results There were significant differences in age,hypertension,new or past stroke,diabetes and hyperhomocysteinemia(HHcy)between the two groups(P<0.05).FIB and D-dimer levels in high-load group were higher than those in low-load group,and PT and APTT were shorter than those in low-load group(P<0.05).Seven predictors of CSVD high load were initially selected.Logistic regression analysis showed that age,hypertension,new or past stroke,FIB,D-dimer and APTT were independent influencing factors for CSVD high load(P<0.05).ROC curve showed that the area under the curve predicted by the Nomograph was 0.802(95%CI:0.743-0.862),and the calibration curve showed that the risk of CSVD high load predicted by the Nomograph was basically consistent with the actual risk situation.The decision curve analysis showed that the Nomogram had a good clinical net benefit.Conclusion Age,hypertension,new or past stroke,FIB,D-dimer and APTT are related to the total CSVD load.The Nomograph prediction model based on the above coagulation indexes has high prediction efficiency and clinical applicability in the prediction of CSVD high load risk,with good differentiation and calibration,and has positive guiding significance for clinical prevention and treatment.