Objective To explore the predictive performance of an artificial intelligence (AI)-assisted ultrasound diagnostic system for tricuspid annular peak systolic velocity (TAPSV) and its value in evaluating right ventricular systolic function. Methods A retrospective analysis was performed on echocardiographic data from 647 patients who underwent cardiac ultrasonography at Shanghai Fourth People’s Hospital Affiliated to Tongji University between May 2023 and Jan. 2024, among whom 75 patients presented with right ventricular systolic dysfunction (defined as TAPSV＜10 cm/s). Patients were randomly assigned to training set (n=517), validation set (n=65), or test set (n=65) in a ratio of 8∶1∶1. The training set was used for AI model parameter learning, the validation set for hyperparameter tuning and overfitting monitoring, and the test set for final independent evaluation. With manually measured TAPSV as the gold standard, the diagnostic efficacy of a deep learning-based AI model for right ventricular systolic dysfunction was evaluated. Results The AI model showed a correlation coefficient of 0.646 (95% confidence interval 0.523-0.740) for TAPSV prediction in the test set, with a mean absolute error (MAE) of 0.94 (relative error: 8.249%) and a root mean square error of 1.08; for right ventricular systolic function diagnosis, the area under the receiver operating characteristic curve was 0.814, with a sensitivity of 0.839 and a specificity of 0.676; and the inter-observer intraclass correlation coefficient and intra-observer intraclass correlation coefficient reached 0.874 and 0.922, respectively, superior to the 0.777 and 0.857 obtained from manual measurements by cardiac ultrasound physicians. Conclusion The AI-assisted ultrasound diagnostic system can predict TAPSV and assist cardiac ultrasound physicians in diagnosing right ventricular systolic dysfunction, showing potential clinical value.