Construction of a risk prediction model for left atrial appendage thrombus/spontaneous echo contrast in patients with nonvalvular atrial fibrillation based on transthoracic echocardiography combined with electrocardiographic parameters and preliminary validation of its efficacy

Left atrial appendage thrombus(LAAT) assessment is a fundamental prerequisite for the formulation of standardized diagnosis and treatment of atrial fibrillation(AF). However, due to technical limitations or poor patient tolerance, conventional modalities such as transesophageal echocardiography and cardiac contrast-enhanced CT leave a large proportion of AF patients without effective LAAT evaluation. This study aims to develop and validate a risk prediction model for LAAT/spontaneous echo contrast(SEC) in patients with first-diagnosed nonvalvular atrial fibrillation(NVAF) based on routine transthoracic echocardiographic and electrocardiographic parameters.

A retrospective cohort study was adopted. A total of 363 patients with newly-diagnosed NVAF admitted to Department of Cardiovascular Medicine of First Affiliated Hospital of Army Medical University from January 2022 to June 2025. Using transesophageal echocardiographic results as the gold standard, they were divided into a LAAT/SEC group(n=87) and a non-LAAT/SEC group(n=276) according to the presence of LAAT/SEC or not. Transthoracic echocardiographic parameters [left atrial diameter(LAD), left ventricular end-diastolic diameter(LVEDD), etc.] and 12-lead electrocardiographic parameters(V1 lead f-wave amplitude, ventricular rate, etc.) were collected from all patients. Multivariate logistic regression analysis was performed to identify independent predictors, and based on the results, a prediction model was constructed, which was visualized as a nomogram. Discriminative ability, calibration performance, and clinical utility of the model were evaluated by the area under the receiver operating characteristic curve(AUC), calibration curve, and decision curve analysis(DCA), and the results were compared with the conventional CHA2DS2-VASc score.

Multivariate logistic regression analysis ultimately identified 4 independent predictors: V1 lead f-wave amplitude ≤0.1 mV, ventricular rate, LVEDD, and LAD. The model based on these predictors demonstrated good predictive performance(AUC=0.831, 95%CI: 0.779 to 0.883). Bootstrap internal validation yielded an AUC value of 0.831(95%CI: 0.781 to 0.882), and a mean of the fold AUCs by 5-fold cross validation of 0.811(95%CI: 0.758 to 0.864). Calibration curve analysis showed good consistency between the frequency of observed results and predicted probabilities(Hosmer-Lemeshow test, P=0.103). DCA revealed a net clinical benefit within a threshold probability range from 0.04 to 0.78. DeLong’s test demonstrated that the predictive performance of the logistic regression model, constructed based on routine transthoracic echocardiographic and ECG parameters, was significantly superior to that of the CHA2DS2-VASc score(AUC=0.574, 95%CI: 0.506 to 0.642, P<0.001). The net reclassification improvement(NRI) was 0.9168 (95%CI: 0.6994 to 1.1341, P<0.001), and the integrated discrimination improvement(IDI) was 0.2379 (95%CI: 0.1817 to 0.2943, P<0.001).

The prediction model for LAAT/SEC in patients with first-diagnosed NVAF constructed in this study demonstrates favorable discriminative ability and clinical applicability, and may serve as a noninvasive auxiliary tool for the rapid identification of high-risk patients and the formulation of individualized anticoagulation strategies in clinical practice.