[Establishment of a predictive nomogram for clinical pregnancy rate in patients with endometriosis undergoing fresh embryo transfer].

Objective: To establish a nomogram model for predicting clinical pregnancy rate in patients with endometriosis undergoing fresh embryo transfer.

Methods: We retrospectively collected the data of 464 endometriosis patients undergoing fresh embryo transfer, who were randomly divided into a training dataset (60%) and a testing dataset (40%). Using univariate analysis, multiple logistic regression analysis, and LASSO regression analysis, we identified the factors associated with the fresh transplantation pregnancy rate in these patients and developed a nomogram model for predicting the clinical pregnancy rate following fresh embryo transfer. We employed an integrated learning approach that combined GBM, XGBOOST, and MLP algorithms for optimization of the model performance through parameter adjustments.

Results: The clinical pregnancy rate following fresh embryo transfer was significantly influenced by female age, Gn initiation dose, number of assisted reproduction cycles, and number of embryos transferred. The variables included in the LASSO model selection included female age, FSH levels, duration and initial dose of Gn usage, number of assisted reproduction cycles, retrieved oocytes, embryos transferred, endometrial thickness on HCG day, and progesterone level on HCG day. The nomogram demonstrated an accuracy of 0.642 (95% CI: 0.605-0.679) in the training dataset and 0.652 (95% CI: 0.600-0.704) in the validation dataset. The predictive ability of the model was further improved using ensemble learning methods and achieved predicative accuracies of 0.725 (95% CI: 0.680-0.770) in the training dataset and 0.718 (95% CI: 0.675-0.761) in the validation dataset.

Conclusions: The established prediction model in this study can help in prediction of clinical pregnancy rates following fresh embryo transfer in patients with endometriosis.