L26/O-136 Conventional morphological grading and AI blastocyst assessment: Prevalence of grading discordance and analysis of pregnancy rates
Human Reproduction · 01.07.2026 · 41 (Supplement_1)
DOI: 10.1093/humrep/deag083.136 ↗Kurz gefasst
Fachartikel in der Zeitschrift Human Reproduction, online erschienen am 08.07.2026. Verfasst von E. Krasnitsky und 5 weiteren Autor:innen.
Abstract (Originalsprache)
Study question What is level of agreement between Gardner grading and AI embryo scores, and what are pregnancy rates of good-Gardner-low-AI blastocysts compared to internal unit results? Summary answer There is a strong alignment between embryologist grading and AI scores. Good-Gardner-low-AI blastocysts were uncommon but demonstrated relatively high pregnancy rates What is known already The Gardner grading system remains the global standard for blastocyst morphological assessment, with high grades strongly associated with increased pregnancy rates. Artificial intelligence (AI) models have been introduced to support embryo selection by providing objective, standardized scores. However, the degree of concordance between conventional morphological grading and AI-based assessment, and the potential implications for clinical outcome, are yet to be fully characterized Study design, size, duration We conducted a retrospective, single-center cohort study following the implementation of the AI-based embryo assessment system iDAScore, consisted of 10,523 blastocysts, created between December 2024 and December 2025. Participants/materials, setting, methods A total of 10,523 blastocysts day 5 and 6 were graded using both Gardner morphology and iDAScore. Blastocysts were categorized into three Gardner quality groups: good (3–4AA, AB, BA n = 4,993), fair (3–4BB, n = 2,446) and poor (3–4AC, CA, BC, CB, CC, n = 3,144); and three iDAScore scoring groups: high (6–9.9, n = 4518), intermediate (3–5.9, n = 3668), and low (1.1–2.9, n = 2339). Discordant subgroups were defined as good-Gardner-low-AI (n = 115) and poor-Gardner-high-AI (n = 144). The Good-Gardner-Low-AI group was further analyzed for pregnancy rate. Main results and the role of chance There is a strong alignment between embryologist grading and AI scores: 70.7% of blastocysts were rated as both good-Gardner and high-AI, while 62.1% were categorized as poor-Gardner and low-AI. Fair-Gardner quality blastocysts showed a diverse range of AI scores, with 52.6% classified as fair-Gardner–intermediate-AI. Outliers were minimal, with 2.3% for good-Gardner–low-AI and 4.6% for poor-Gardner–high-AI blastocysts. 95 blastocysts from the good-Gardner-low-AI subgroup were further cryopreserved, 21 blastocysts were thawed and transferred, 9 of them resulted in pregnancy (42.8%). Similar pregnancy rates for thawed blastocysts graded by the same two methods during that same period of time were achieved at our unit (46%, P = 0.9). Limitations, reasons for caution The study cohort was limited by the small number of transfers in the examined blastocysts subgroup. Image quality variations and missing data have also contributed to the reduction in the study group. Missing confounders, including endometrial receptivity and underlying infertility diagnoses, may influence outcomes. Wider implications of the findings This research proposal addresses a gap by evaluating the pregnancy rates of grading mismatches between traditional Gardner criteria and AI-scores. While the high-Gardner-low-AI blastocysts are uncommon, they have demonstrated significant pregnancy potential in patients, information which may render valuable for tailor-made treatment. Trial registration number No
Zitieren
APA
Krasnitsky, E., Michailov, Y., Yaacobi-Artzi, S., Shvitzer, G., Tzarfati, O., & Shavit, T. (2026). L26/O-136 Conventional morphological grading and AI blastocyst assessment: Prevalence of grading discordance and analysis of pregnancy rates. Human Reproduction, 41(Supplement_1). https://doi.org/10.1093/humrep/deag083.136
Vancouver
Krasnitsky E, Michailov Y, Yaacobi-Artzi S, Shvitzer G, Tzarfati O, Shavit T. L26/O-136 Conventional morphological grading and AI blastocyst assessment: Prevalence of grading discordance and analysis of pregnancy rates. Human Reproduction. 2026;41(Supplement_1). doi:10.1093/humrep/deag083.136
BibTeX
@article{Krasnitsky2026,
author = {Krasnitsky, E. and Michailov, Y. and Yaacobi-Artzi, S. and Shvitzer, G. and Tzarfati, O. and Shavit, T.},
title = {L26/O-136 Conventional morphological grading and AI blastocyst assessment: Prevalence of grading discordance and analysis of pregnancy rates},
journal = {Human Reproduction},
year = {2026},
volume = {41},
number = {Supplement_1},
doi = {10.1093/humrep/deag083.136}
}
EUVIMED