Prognostic value of a machine learning algorithm based on multimodal echocardiographic parameters in patients with bicuspid aortic valve
Liu Jing
Zhang Junyue
Jia Juan
Cheng Shuai
Abstract：Objective To evaluate the prognostic value of machine learning algorithms based on multimodal echocardiographic parameters in patients with bicuspid aortic valve(BAV).Methods The clinical data of 92 patients with BAV treated at Zhengzhou No.7 Peoples's hospital from June 2019 to June 2023 were retrospectively analyzed.According to the 1-year follow-up results,patients were divided into good prognosis group and poor prognosis group.Logistic regression analysis was used to identify risk factors for poor prognosis in BAV patients.Logistic regression,random forest,and support vector machine(SVM)models were constructed based on multimodal echocardiographic parameters.The receiver operating characteristic(ROC)curves of the three models were plotted to evaluate the their predictive performance.Results The proportion of type-1(right-left coronary cusp fusion)BAV,left ventricular mass index(LVMI),ascending aortic diameter,and peak systolic velocity of the aortic valve were higher in the poor prognosis group than in the good prognosis group,while the aortic valve orifice area was lower(P＜0.05).Logistic regression analysis showed that aortic valve morphology(OR=5.018,95%CI:2.734～7.302,P=0.013),LVMI(OR=5.114,95%CI:1.225～9.003,P=0.011),ascending aortic diameter(OR=6.190,95%CI:1.354～11.027,P=0.001),aortic valve orifice area(OR=4.933,95%CI:1.754～9.113,P=0.006),and peak systolic velocity of the aortic valve(OR=5.485,95%CI:1.586～10.384,P=0.002)were independent risk factors for poor prognosis in BAV patients.Among the models constructed with these parameters,the SVM model had the highest AUC(0.914,95%CI:0.851～0.977,P＜0.001),compared to the logistic regression model(AUC=0.845,95%CI:0.763～0.927,P=0.002)and the random forest model(AUC=0.826,95%CI:0.731～0.921,P=0.005).Conclusion The prognosis of BAV patients is closely related to valve morphology and hemodynamic parameters.The SVM model demonstrated the best predictive performance in the comprehensive analysis of multimodal echocardiographic parameters.It is recommended the dynamic monitoring and intervention be implemented for high-risk factors(type-1 BAV,ascending aortic dilation,increased LVMI,valve stenosis,and accelerated blood flow)to improve patient outcomes.
Keywords：Bicuspid aortic valveEchocardiographyMachine learning algorithmPrognosis prediction
Publication Date：2025-09-20
Online Publishing Date：2025-11-04(First online date of this platform, not the publication date of the document)
Pages：5( 1068-1072 )