Desarrollo de un Modelo de Riesgo para Baja Recuperación de Ovocitos en Pacientes de FIV de Primer Ciclo con Reserva...

Low oocyte retrieval (LOR), defined as oocytes retrieved from <50% of mature follicles, significantly impacts IVF success, especially in patients with diminished ovarian reserve (DOR). Predictive models tailored for this population are limited. This retrospective cohort study developed a predictive nomogram for LOR risk in first-cycle IVF/ICSI patients with DOR. Data from 2,594 eligible patients were analyzed. LOR was defined as oocyte retrieval rate <50% (oocytes/follicles ≥14mm on trigger day). Least absolute shrinkage and selection operator (LASSO) regression identified predictors from clinical/endocrine parameters. A multivariate logistic model was built, visualized as a nomogram, and internally validated (bootstrap resampling). Performance was assessed via AUC, calibration, and decision curve analysis. Eight independent predictors were identified: Age, basal FSH, AFC, AMH, LH on hCG day, Progesterone (P) on hCG day, number of large follicles (≥14mm) on hCG day, and trigger-to-retrieval time (TR-OPU). The model demonstrated moderate yet acceptable discrimination (AUC = 0.721, 95% CI: 0.678-0.772) and calibration. TR-OPU was significantly shorter in the LOR group (34.85 vs. 35.05 hours, p<0.001). DCA confirmed clinical utility. This study establishes a clinical-endocrine nomogram predicting LOR risk in first-cycle DOR patients. Incorporating key factors like TR-OPU, ovarian reserve markers (AMH, AFC, bFSH), and trigger-day hormones, it may help stratify risk but requires external validation before clinical application.