L26/P-210 Biological heterogeneity among euploid embryos: Artificial Intelligence (AI)-based prioritization for live birth and reduced odds of miscarriage
Human Reproduction · 01.07.2026 · 41 (Supplement_1)
DOI: 10.1093/humrep/deag083.546 ↗Kurz gefasst
Fachartikel in der Zeitschrift Human Reproduction, online erschienen am 08.07.2026. Verfasst von S. Kahraman und 8 weiteren Autor:innen.
Abstract (Originalsprache)
Study question In patients with multiple euploid embryos available, can an AI embryo quality score identify embryos with the highest live birth potential and lower miscarriage risk? Summary answer AI-derived embryo scores independently predicted euploidy, live birth, and miscarriage risk, enabling objective prioritization of euploid embryos for transfer. What is known already Preimplantation genetic testing for aneuploidy (PGT-A) has reduced aneuploid embryo transfer, yet implantation failure and miscarriage still occur in a substantial proportion of euploid transfers. When multiple euploid embryos are available, IVF professionals must decide which embryo to transfer first, with limited additional developmental information beyond conventional embryo grading which provides limited discriminatory power within euploid cohorts. As a result, euploid embryo prioritization continues to rely largely on embryologist experience rather than objective evidence. AI embryo assessment is increasingly integrating into routine IVF workflows, creating the opportunity to evaluate its role in euploid embryo prioritization. Study design, size, duration Retrospective single-center cohort study including autologous PGT-tested blastocysts and single euploid embryo transfers performed between 2021–2025. Patients with at least two available euploid embryos were included. Embryo and transfer analyses evaluated the associations between an AI-derived embryo quality score (range 0–10) and euploidy, live birth (LB), and miscarriage outcomes. Participants/materials, setting, methods 1,047 PGT-tested blastocysts, including 727 euploid and 320 aneuploid embryos, from 269 patients were analyzed. Euploidy was assessed at the embryo level using multivariable logistic regression adjusted for maternal age. LB analysis included 177 single euploid transfers from 143 patients. Miscarriage analysis was conditional to clinical pregnancy and included 179 implanted euploid transfers from 163 patients. Transfer-level analyses compared embryos within the same cohort. Model discrimination was assessed using AUC adjusted for maternal age. Main results and the role of chance Mean patient age was 33.2 years, mean BMI 24.4, and mean number of embryos per patient 4.6. Among PGT-tested blastocysts, the euploidy rate was 69.4% (727/1047). In euploid single embryo transfers, live birth and miscarriage rates were 70.6% (125/177) and 5.6% (10/179), respectively. Higher AI quality scores were associated with increased odds of euploidy; each 1-point increase in the score corresponded to a 14% increase in odds of euploidy (p < 0.001). Mean AI scores were higher in euploid embryos compared with aneuploid embryos (7.80 vs 6.99), with an AUC of 0.62 for euploidy prediction. Within euploid transfers, transfer of the top-ranked embryo by AI was associated with higher LB odds (OR 2.31, p < 0.05), corresponding to an increase in LB probability of 17%. LB rates were 72.5% (137/189) for top-ranked versus 52.5% (32/61) for non-top ranked embryos. Furthermore, the AI model was predictive of LB with an AUC of 0.64 with higher mean AI scores in LB embryos than in non–LB (8.04 vs 7.65). Among clinically implanted euploid embryos, AI top-ranked embryos had 76% lower odds of miscarriage (OR: 0.24, p < 0.05). Mean AI scores were lower in miscarriages than non-miscarriages (7.44 vs 8.22), with an AUC of 0.60 (p < 0.05). Limitations, reasons for caution Single-center retrospective design and enrichment for PGT-tested blastocysts may limit generalizability and inflate baseline euploidy rates. High success rates among euploid transfers reduce power to detect smaller effects. Wider implications of the findings Findings support biological heterogeneity among euploid embryos and indicate that AI-derived embryo scores encode post-ploidy developmental competence. This information may enable objective prioritization within euploid cohorts, improving live birth outcomes while reducing the odds of miscarriage. Trial registration number No
Zitieren
APA
Kahraman, S., Colakoglu, Y. K., Cetinkaya, C. P., Yelke, H., Kelley, K., Wiemer, K., Popovic, M., Ben-Meir, A., & Pini, P. (2026). L26/P-210 Biological heterogeneity among euploid embryos: Artificial Intelligence (AI)-based prioritization for live birth and reduced odds of miscarriage. Human Reproduction, 41(Supplement_1). https://doi.org/10.1093/humrep/deag083.546
Vancouver
Kahraman S, Colakoglu YK, Cetinkaya CP, Yelke H, Kelley K, Wiemer K, et al. L26/P-210 Biological heterogeneity among euploid embryos: Artificial Intelligence (AI)-based prioritization for live birth and reduced odds of miscarriage. Human Reproduction. 2026;41(Supplement_1). doi:10.1093/humrep/deag083.546
BibTeX
@article{Kahraman2026,
author = {Kahraman, S. and Colakoglu, Y. Kumtepe and Cetinkaya, C. Pirkevi and Yelke, H.K. and Kelley, K. and Wiemer, K. and Popovic, M. and Ben-Meir, A. and Pini, P.},
title = {L26/P-210 Biological heterogeneity among euploid embryos: Artificial Intelligence (AI)-based prioritization for live birth and reduced odds of miscarriage},
journal = {Human Reproduction},
year = {2026},
volume = {41},
number = {Supplement_1},
doi = {10.1093/humrep/deag083.546}
}
EUVIMED