Objective ：To construct a nomogram model for predicting pregnancy outcomes of frozen-thawed embryo transfer based on ultrasound parameters on the day of endometrial transformation. Methods :78 patients who planned to undergo frozen embryo transfer in our hospital from January 2021 to July 2023 were selected as the research objects. All patients underwent vaginal ultrasound examination on the day of endometrial transformation, and the endometrial classification, thickness, endometrial volume, subendometrial area flow index (FI), vascularization index (VI), and vascularization flow index (VFI) were observed. Uterine artery resistance index (RI), pulsatility index (PI), uterine artery blood flow peak systolic velocity/end diastolic velocity (S/D). All patients were divided into pregnancy (+) group and pregnancy (-) group according to pregnancy status. The clinical characteristics and ultrasound parameters of the two groups were compared, and multivariate Logistic regression analysis was performed. Radiomics features were extracted and screened to construct a nomogram model and predict its diagnostic efficacy.Results : There were significant differences in clinical and ultrasound parameters such as age, infertility type, basal FSH and embryo type between the pregnancy (+) group and the pregnancy (-) group (P < 0.05). According to the above comparison of clinical and ultrasound parameters, it was found that the influencing factors of pregnancy outcome of frozen embryo transfer cycle were age, infertility type, basal FSH, embryo type and other clinical parameters. The influencing factors were classified according to age (< 35, ≥35), type of infertility (primary infertility, secondary infertility), basal FSH (< 10, ≥10), and type of embryo (cleavage embryo, blastocyst). The influencing factors for high pregnancy rate were age < 35 years, primary infertility, basal FSH≥10 miU /ml, and blastocyst. The prediction model was established based on the above factors. The radiomics features with strong repeatability were further optimized and streamlined, and the 9 best radiomics features were finally obtained. The Rad-score of patients with (+) pregnancy in the validation set was significantly higher than that of patients with (-) pregnancy (0.36 vs.-0.15, P=0.007). The Rad-score of patients with (+) pregnancy in the training set was significantly higher than that of patients with (-) pregnancy (0.32 vs.0.04, P < 0.001). The cut-off value of Rad-score for evaluating pregnancy was 0.18. ROC curve analysis showed that in the validation cohort, the Rad-score had an AUC of 0.76 (95%CI: 0.56-0.93), a specificity of 0.76, a sensitivity of 0.66, a negative predictive value of 0.66, and a positive predictive value of 0.82. In the training set, the Rad-score had an AUC of 0.73 (95%CI: 0.62-0.82), a specificity of 0.61, a sensitivity of 0.76, a negative predictive value of 0.66, and a positive predictive value of 0.72. Based on the above multivariate analysis, the independent clinical factors were determined as infertility type, age, embryo type and basal FSH. After multivariate regression analysis with radiomics signature, the results showed that the independent predictors were age, embryo type and radiomics signature, and the clinical-ultrasound radiomics combined prediction model was constructed. The ROC curve of the clinical features, radiomics signature and nomogram combined prediction model in the validation set and training set showed that the nomogram combined prediction model had good prediction efficiency. In the validation cohort, the area under the ROC curve of the clinical feature model was 0.60, while that of the nomogram combined prediction model was 0.81. In the training set, the area under the ROC curve of the clinical feature model was 0.66, while that of the nomogram combined prediction model was 0.82. In the validation set and the training set, the nomogram prediction model showed good consistency on the calibration curve for the observed and predicted pregnancy values. The clinical application efficacy of the nomogram prediction model was analyzed by the decision curve, and the results showed that the pregnancy prediction probability was in the range of 0.13-0.72. The nomogram prediction model has a better net benefit than the clinical characteristics model in predicting the pregnancy rate of frozen-embryo transfer cycle.Conclusion :The nomogram prediction model based on ultrasound radiomics can be used to analyze the endometrium on the endometrial transformation day before embryo transfer, predict the pregnancy rate, and formulate a personalized embryo transfer plan, so as to improve the pregnancy outcome of assisted reproductive technology.