Study on predicting vascular embolism events in cardiac myxoma with binary logistic regression model based on clinical and ultrasound characteristics
Objective To explore the risk factors for vascular embolism (VE) in patients with cardiac myxoma (CM) based on clinical and ultrasound characteristics,and to construct a binary logistic regression model for predicting VE events in patients with CM. Methods The retrospective screening was conducted on 120 patients who were confirmed to have CM by pathology in Dezhou Hospital,Qilu Hospital,Shandong University from June 2016 to July 2023. Based on whether these patients experienced VE before surgery,they were divided into the VE group and the control group. Univariate and multivariate analyses on the clinical and echocardiographic characteristics of two groups were performed. A binary logistic regression model was established. The predictive efficacy and fitting ability of the model were assessed by using receiver operating characteristic (ROC) curve and Hosmer-Lemeshow test. Results The results of univariate analysis showed that age,smoking history,tumor morphology and structure were the factors affecting the occurrence of VE in CM patients;Multivariate analysis showed that smoking history,irregular tumor morphology,and lax tumor structure were independent risk factors for VE in CM patients. Conclusion A binary logistic risk prediction model established based on clinical and ultrasound characteristics has good predictive efficacy and fitting ability,which can provide the certain reference basis for the prevention of VE in the patients with CM.
echocardiographycardiac myxomavascular embolismprediction model
万方数据
