Practical prediction model for day 4 embryonic development potential: integration of morphology and AI scoring

Area of Science:

• Reproductive Medicine
• Embryology
• In Vitro Fertilization (IVF)

Background:

• Accurate prediction of embryonic developmental potential is crucial for optimizing In Vitro Fertilization (IVF) outcomes.
• Current methods for assessing embryo viability have limitations in predicting success rates.

Purpose of the Study:

• To develop and validate a practical predictive model for day 4 embryonic developmental potential.
• To identify key factors influencing transferable blastocyst formation.
• To evaluate the clinical applicability of the predictive model.

Main Methods:

• Logistic regression analysis was used to identify independent predictors of blastocyst formation.
• Three models were compared based on day 3 morphology, iDAScore, and day 4 morphology.
• Receiver operating characteristic (ROC) analysis and calibration analyses were performed to evaluate model performance.

Main Results:

• Degree of compaction, iDAScore, and day 3 cell numbers were identified as independent predictors.
• The comprehensive model (Model 3) achieved an AUC of 0.841 in the training and 0.828 in the validation cohort for blastocyst formation.
• The nomogram demonstrated strong calibration and accurately predicted high-quality blastocyst formation, implantation, and live birth rates.

Conclusions:

• Day 4 compaction, iDAScore, and day 3 cell numbers are key predictors of transferable blastocyst formation.
• The developed nomogram serves as an effective tool for predicting embryonic developmental potential and IVF success.
• The model shows strong performance across multiple clinical outcomes, enhancing its applicability in fertility treatments.