Abstract
This study aimed to construct and validate a nomogram for predicting natural pregnancy after hysteroscopy and laparoscopy in patients with tubal infertility and pelvic endometriosis, providing a basis for precise clinical evaluation and personalized treatment. This retrospective observational study included 523 patients with tubal infertility complicated with pelvic endometriosis treated at Beijing Shijitan Hospital from January 2018 to January 2023. Clinical data were retrospectively collected and randomly divided into a model group (n = 366) and a validation group (n = 157) in a 7:3 ratio. Univariate analysis, least absolute shrinkage and selection operator regression, and multivariate logistic regression were used to identify independent predictors for the nomogram. The model's performance was evaluated using receiver operating characteristic curves, calibration plots, and decision curve analysis. Through univariate analysis, least absolute shrinkage and selection operator algorithm screening, and multivariate logistic regression, age (odds ratio [OR] = 1.294), infertility duration (OR = 1.065), American Society for Reproductive Medicine stage (OR = 1.773), and cancer antigen 125 (OR = 2.431) were identified as independent influencing factors. A prediction model was constructed based on these factors. The receiver operating characteristic curve showed AUCs of 0.771 (95% CI = 0.718-0.823) in the model group and 0.731 (95% CI = 0.639-0.823) in the validation group. The Hosmer-Lemeshow test indicated no significant difference between predicted and actual pregnancy probabilities (P > .05). Decision curves showed maximum net benefits at threshold probabilities of 0.19 to 0.65 (model group) and 0.12 to 0.68 (validation group), indicating good clinical efficacy. The 4 factors are independent predictors, and the nomogram based on them demonstrates good predictive value, providing a useful basis for clinical prognosis, intervention, and individualized treatment planning. However, as this is a retrospective single-center study, the model requires external validation in multicenter cohorts to confirm its generalizability and clinical applicability.