Bridging the gap between embryo euploidy, pregnancy potential and morphology using artificial intelligence for ploidy...

Area of Science:

• Reproductive medicine and artificial intelligence.
• Embryology and in vitro fertilization (IVF).

Background:

• Accurate embryo selection is crucial for successful IVF outcomes.
• Current methods for assessing embryo ploidy can be invasive or lack precision.

Purpose of the Study:

• To develop and validate a non-invasive artificial intelligence (AI) model using time-lapse imaging (TLI) data to predict embryo ploidy.
• To assess the AI model's ability to support embryo selection in IVF.

Main Methods:

• A retrospective multicentre study trained an AI model on TLI videos from 4822 embryos.
• The model generated a continuous score indicating euploidy likelihood.
• External validation was performed on 706 embryos from an independent cohort.

Main Results:

• Higher AI ploidy scores correlated significantly with euploidy and fetal heartbeat presence.
• Euploid embryos exhibited higher AI scores than aneuploid ones (51.3 vs. 43.44).
• The AI model demonstrated generalizable performance across clinics, outperforming existing predictors.

Conclusions:

• A validated, non-invasive AI model accurately predicts embryo ploidy from TLI data.
• The AI model enhances embryo selection by adding value to morphological assessment.
• This technology is particularly beneficial when preimplantation genetic testing for aneuploidies is not feasible.