A risk prediction model for lower limb venous thrombosis after cerebral hemorrhage surgery based on LASSO-logistic regression

HAO Guangzhi; DONG Yushu; ZHANG Bingying; SUN Linlin; GAO Yang; XIONG Jian

Abstract：Objective To construct a clinical prediction model on the risk factors of postoperative venous thrombosis in patients with intracerebral hemorrhage(ICH)based on LASSO-logistic regression.Methods A total of 768 patients with spontaneous cerebral hemorrhage admitted to the General Hospital of the Northern Theater Command from February 2023 to April 2024 were selected and randomly divided into the training set and the verification set according to the 1:1 ratio.Lasso regression and univariate logistic regression were used in the training set of cases to search for suspected risk factors leading to venous thrombosis after intracerebral hemorrhage.Independent risk factors were determined by multivariate logistic regression and a prediction model on such rick factors was constructed.Nomogram was drawn for visual display,the area under ROC curve was calculated,calibration chart was drawn to evaluate the constructed model,and clinical application value of the model was evaluated by clinical decision curve analysis(DCA).Results The incidence of post-ICH venous thrombosis was correlated with age,diabetes,GCS score and blood fibrinogen content(P<0.05),rather than blood pressure,sex,drinking,smoking,and operation mode(P>0.05).The area under ROC curve of the model was 0.840(95%CI:0.773-0.907).The good consistency of calibration curves and the Hosmer-Lemeshow goodness of fit test(χ2=9.596,P=0.384)showed that the model had good differentiation and calibration.The threshold probability in DCA was between 0.02 and 0.80,which indicated that the model had a good net benefit.Conclusion Advanced age(>60),diabetes mellitus,low GCS score,and fibrinogen(>4 g/L)are independent influencing factors for the occurrence of lower extremity venous thrombosis after spontaneous intracerebral hemorrhage.The prediction model based on these factors has good clinical application value.