Development and validation of a predictive model for early blastocyst formation on day 4 post-fertilization

Objective To develop and validate a predictive model for early blastocyst formation on Day 4 post-fertilization using clinically accessible parameters. Methods This retrospective cohort of 2,557 patients scheduled for fresh Day 4 embryo transfer was randomly divided into training (n=1,789) and validation (n=768) sets at Xingtai MeiHe Reproductive and Genetic Hospital from January 2021 to December 2025. Early blastocyst formation occurred in 44.9% (training) and 44.1% (validation) of cases. Multivariable logistic regression and nomogram modeling were employed, with model performance assessed via AUC, calibration curves, and decision curve analysis. Results After multivariable adjustment, the number of Day 3 embryos with 10 blastomeres (aOR=1.668, 95% CI: 1.374~2.025, P0.001) and those with iDAScore 4.8 (aOR=1.369, 95% CI: 1.190 ~ 1.575, P0.001) were the only independent predictors of Day 4 early blastocyst formation, while 15 covariates including ovarian reserve markers and sperm parameters showed no significant association (P0.05). The nomogram integrating Day 3 embryos with 10 blastomeres and iDAScore4.8 demonstrated robust discrimination (training AUC = 0.782, 95% CI: 0.738 ~ 0.826; validation AUC = 0.773, 95% CI: 0.705 ~ 0.841) with sensitivity of 65.4~65.9% and specificity of 79.6 ~ 81.7%. Calibration curves indicated minimal prediction deviation, and decision curve analysis confirmed clinical net benefits across 18~80% threshold probabilities in the training cohort and 20~93% in the validation cohort. Clinically, early blastocyst formation was associated with superior embryo quality parameters (P0.001), higher single embryo transfer rates (50.90% vs. 13.96%, P0.001), and increased clinical (67.65% vs. 59.32%, P0.001) and ongoing pregnancy rates (59.59% vs. 52.89%, P = 0.006), alongside a lower multiple pregnancy rate (22.50% vs. 31.93%, P0.001). Multivariable regression confirmed early blastocyst formation as an independent predictor of pregnancy success (clinical pregnancy aOR=1.675, 95% CI: 1.287 ~ 2.180, P0.001; ongoing pregnancy aOR=1.612, 95% CI: 1.249 ~ 2.080, P0.001). Conclusion The validated nomogram accurately predicts Day 4 early blastocyst formation using Day 3 embryo parameters. The presence of early blastocyst formation on Day 4 might indicate suitability for selective Day 4 embryo transfer.