Machine-Learning Prediction of Pregnancy Outcomes After Assisted Reproductive Treatments

The goal of this observational study is to develop and validate a machine-learning model to predict pregnancy outcomes in women with a positive pregnancy test following assisted reproductive treatment (ART), with the primary analysis focusing on in vitro fertilization (IVF) and frozen embryo transfer (FET). The study will also include an exploratory analysis of pregnancies following intrauterine insemination (IUI). The main questions addressed by the study are: Can early blood levels of beta-human chorionic gonadotropin (beta-hCG) predict the likelihood of live birth in an IVF or FET cycle? Can early beta-hCG levels help identify pregnancies at increased risk of biochemical pregnancy or early pregnancy loss? Participants will have clinical and treatment information, including beta-hCG measurements and pregnancy outcomes, collected from medical records or during prospective follow-up. No changes to participants' medical treatment will be made as part of the study.

Inclusion Criteria: * Women who underwent ART procedures (IUI, IVF/ICSI, or FET) using homologous oocytes. * Availability of at least one positive early serum beta-HCG measurement after aspiration, FET, or insemination (beta-HCG \>10 mIU/mL). * Availability of the pregnancy outcome. * Availability of relevant patient characteristics and clinical variables required for predictive modeling. * For embryo-transfer cycles, availability of information on embryo developmental stage and number of embryos transferred where applicable. * For the prospective validation cohort, written informed consent. Exclusion Criteria * Missing beta-HCG measurements critical for model input. * Missing key demographic or clinical variables required for predictive modeling. * Missing ART treatment protocol where required for the model. * Missing pregnancy outcomes. * Multiple pregnancy, hydatidiform mole, heterotopic pregnancy, and ectopic pregnancy will be excluded from the primary live-birth analysis but will be described separately and, where sample size permits, evaluated as secondary adverse pregnancy outcomes. * Non-consent for prospective participation in validation.