Decoding embryo development: the effect of clinical variables in morphokinetics and artificial intelligence quality...

Do morphokinetic events (MKS) and patient parameters affect resulting embryo quality scores, and how does this relate to pregnancy outcomes as assessed by a time-lapse incubator using an artificial intelligence (AI) embryologist support tool?

Retrospective study analysing data from 6024 embryos retrieved over 1636 cycles. The dataset comprised 3778 donor oocytes and 2246 autologous oocytes. Additionally, 3309 biopsied embryos were included in the PGT-A analysis. Outcome data were derived from 1355 transferred embryos. All embryos were assessed using a time lapse-based AI system (CHLOE EQ™), which assigns an embryo quality score from 0 to 1 based upon established morphokinetic benchmarks. The AI embryo quality score was assigned on day 5 of embryo development. Analysed patient parameters were patient age, fresh or frozen oocyte status and use of own or donor oocytes. Twenty-three MKS were analysed. The effect of morphokinetics on the incidence of ploidy was also assessed.

Clinical outcomes are affected by embryo quality and MKS as detected by an AI software. Time to expanded blastocyst (tEB) was the morphokinetic parameter with the strongest correlation coefficient with embryo quality score (-0.816). As the patient's age increases by 1 unit, embryo quality score decreases significantly and time to achieve expanded blastocyst increases by 0.47 h. Similar trends were observed with frozen oocytes, which showed a 2.1 h increase in tEB compared with fresh oocytes. Autologous oocytes were associated with a 6.08 h longer tEB compared with donor oocytes. Additionally, euploid embryos reached tEB 4.72 h earlier than aneuploid embryos. For a one unit increase in embryo quality score, the odds of achieving clinical pregnancy increased by 21.7% and the odds of achieving ongoing pregnancy or live birth increased by 18.5%. Oocyte sources had an effect on miscarriage rates; the use of frozen oocytes resulted in higher miscarriage rates than observed when fresh oocytes were used.

AI can successfully evaluate embryo quality and can assist embryologists in decision making. Furthermore, this AI model can delineate the effect of various clinical factors on resulting outcomes.