Construction and Validation of a Risk Prediction Model for Disease Pro-gression of Posterior Circulation Cerebral Infarction Patients Based on Thromboelastography

ZHANG Chongsheng; XIE Juan; WANG Qianyou

doi:10.3969/j.issn.1002-3429.2024.19.011

Abstract：Objective To construct a risk prediction model for the progression of posterior circulation cerebral infarc-tion(PCCI)patients based on thromboelastogram,aiming to provide the optimal predictive tool for clinical practice.Methods A total of 120 patients with PCCI from May 2021 to April 2023 were selected as the modeling population,and another 80 patients with PCCI treated during the same period were selected as the validation population.The disease progression,clinical data and thromboelastic parameters of the patients were analyzed,and the characteristic variables of disease progression of pa-tients with PCCI were preliminarily screened by LASSO.Logistic regression was used to analyze the influencing factors of dis-ease progression.R software was used to construct the nomogram prediction model,and the differentiation,validity and accu-racy of the model were evaluated by receiver operating characteristic(ROC)curve,clinical decision curve and calibration curve,respectively.Results LASSO regression analysis showed that when the penalty coefficient λ=0.122,the model had good performance with the least influencing factors.Six predictive variables were finally selected,including the history of atrial fibrillation,the score of the National Institutes of Health Stroke Scale(NIHSS)at admission,the reaction time of throm-boelastic parameters,the coagulation time,the maximum amplitude,and the coagulation angle.Logistic regression analysis showed that NIHSS score,history of atrial fibrillation,reaction time of thromboelastic parameters,coagulation time,the maxi-mum amplitude and coagulation angle were all influencing factors for disease progression in patients with PCCI(P＜0.01).According to the above influencing factors,the risk prediction model of disease progression in patients with PCCI was estab-lished.In the modeling and verification population,the area under the ROC curve(AUC)of the nomogram model was0.875(95%CI:0.813,0.937)and 0.914(95%CI:0.851,0.976),respectively,indicating that the model had good differentia-tion.The calibration curve shows that the predicted value of the nomogram model was highly correlated with the actual observa-tion results in the modeling and verification population,indicating good accuracy.The clinical decision curve showed that the net benefit value of the nomogram model was better in the modeling and verification population,suggesting that the model had good predictive efficiency.Conclusion History of atrial fibrillation,NIHSS score at admission,reaction time of thromboelas-tic parameters,coagulation time,maximum amplitude,and coagulation angle were all influencing factors for the progression of patients with PCCI.The nomogram model built based on the above influencing factors had good risk prediction efficiency.