Does embryo categorization by existing artificial intelligence, morphokinetic or morphological embryo selection...

Area of Science:

• Reproductive Medicine
• Embryology
• Artificial Intelligence in Medicine

Background:

• Preimplantation genetic testing for aneuploidies (PGT-A) is crucial for optimizing IVF outcomes.
• Embryo selection models aim to improve implantation rates but their correlation with euploidy is debated.
• Existing models include artificial intelligence (AI), morphokinetic, and morphological grading systems.

Purpose of the Study:

• To investigate the correlation between existing embryo scoring models and blastocyst euploidy.
• To assess if AI, morphokinetic (KIDScore Day 5), and morphological (Gardner criteria) models align with euploidy rates across different maternal age groups.

Main Methods:

• Retrospective study of 3573 blastocysts from 834 patients undergoing PGT-A.
• Patients stratified into five maternal age groups.
• Correlation analysis between euploidy rates and iDAScore (AI), KIDScore Day 5 (morphokinetic), and Gardner criteria (morphological).

Main Results:

• Significant correlation found between euploidy rates and iDAScore across all age groups (P < 0.001).
• Significant correlation also observed with KIDScore Day 5 and Gardner criteria in most age groups (P < 0.0001).
• Multivariate logistic regression confirmed higher scores in all models significantly correlated with euploidy (P < 0.0001).

Conclusions:

• Current blastocyst scoring models demonstrate a correlation with ploidy status.
• These models, while not diagnostic for euploidy, can support decisions on which embryos to biopsy for PGT-A.
• Utilizing these models may help reduce patient costs associated with PGT-A by optimizing biopsy selection.