L26/P-229 Artificial intelligence (AI)-derived oocyte ploidy score predicts PGT-A outcome beyond conventional morphology, with increasing value at advanced maternal age
Human Reproduction · 01.07.2026 · 41 (Supplement_1)
DOI: 10.1093/humrep/deag083.565 ↗Kurz gefasst
Fachartikel in der Zeitschrift Human Reproduction, online erschienen am 08.07.2026. Verfasst von Y. Franco Iriarte und 14 weiteren Autor:innen.
Abstract (Originalsprache)
Study question Can PGT-A outcomes be predicted by an AI-derived oocyte ploidy score or visual oocyte dysmorphisms, and how does maternal age influence these predictions? Summary answer MAGENTATM Ploidy-AI score significantly predicts PGT-A outcome, with stronger performance at advanced maternal age (AMA), while conventional oocyte dysmorphic categories show no consistent association. What is known already Maternal age is widely used as a non-invasive indicator of embryo ploidy, with AMA associated with greater aneuploidy occurrence; however, it does not discriminate between oocytes within the same patient’s cohort. Visual oocyte dysmorphic features are recorded in some laboratories, and their association with embryo developmental competence and ploidy has been explored. Yet, the results remain inconsistent, and no consensus supports their clinical value. AI-based analysis of metaphase II (MII) oocyte images has been associated with embryo developmental competence and more recently ploidy status, although evidence for predicting euploidy across different maternal ages from the oocyte stage is still emerging. Study design, size, duration This single-center, retrospective study conducted between 2022 and 2023, analyzed 2,007 mature MII oocytes from 322 patients (ages 18-48).[EM1] Of 755 blastocysts with PGT-A results, 266 were euploid, 25 low-level mosaic, 40 high-level mosaic, and 424 aneuploid. Euploid and low-level mosaics were grouped together, and high-level mosaics and aneuploids were grouped for analyses. Analyses were performed at both oocyte and biopsied embryo levels, with failure to reach blastocyst considered a negative outcome. Participants/materials, setting, methods Oocyte images were acquired immediately after ICSI using a Geri time-lapse incubator. Morphological assessments were performed by senior embryologists, recording the presence of specific oocyte dysmorphisms: altered zona pellucida (ZP, n = 77), perivitelline space abnormalities (PVS, n = 411), cytoplasmic granularity (n = 257), and smooth endoplasmic reticulum (SER, n = 37). The same images were analyzed using MAGENTA™ Ploidy-AI, which generated a novel ploidy score (0-10). Associations with PGT-A outcome were evaluated using χ² tests and generalized estimating equation (GEE) models. Main results and the role of chance At both oocyte and embryo levels, MAGENTA™ Ploidy-AI score was strongly associated with PGT-A outcomes (global p < 0.001). When stratified into score intervals (0-<2.5; 2.5-<5.0; 5.0-<7.5; 7.5-10), the proportion of euploid embryos increased progressively with higher Ploidy-AI scores. Likewise, aneuploidy rates decreased between the lowest and highest score groups. At the embryo level, euploidyrate increased from 17% in the loweest score group to 48% in the highest group while aneuploidy rate conversely decreased from 76% to 39%. Pairwise comparisons of euploidy rates (both oocyte/embryo level) showed significant differences between the lowest and successively higher score groups (p < 0.001), whereas differences among intermediate and high score intervals were largely non-significant. No significant differences were observed for high- or low-level mosaic proportions. None of the evaluated dysmorphic categories (cytoplasmic granularity, SER, PVS, ZP) were associated with PGT-A outcomes (p > 0.05). At the oocyte level, GEE models showed that each unit increase in the Ploidy-AI score increased the odds of euploidy by 30.8% (OR 1.31, 95% CI 1.21-1.41, p < 0.001). Notably, the model’s predictive power strengthened with age, increasing by 2.7% per year (OR 1.03, 95% CI 1.01-1.05, p < 0.01), reflecting enhanced discriminatory capacity in older patients. Limitations, reasons for caution This study is limited by its retrospective and single-center nature. Oocyte dysmorphisms were recorded as presence or absence only, without quantitative grading – greater details should be investigated in future research. Mosaic embryos were not the primary outcome and were limited in number, warranting cautious interpretation and further dedicated studies. Wider implications of the findings MAGENTA Ploidy-AI score provides a more reliable ploidy biomarker than traditional morphological oocyte observations, especially in AMA – enabling insights from the earliest stage. The stability of the predictive performance when mosaic embryos were included motivates further investigation of AI-based approaches in larger, multicenter cohorts with greater mosaic representation. Trial registration number No
Zitieren
APA
Iriarte, Y. F., Villa, G. B., Milla, A. V., Burgos, L. B., Borras, F. S., Aguilar, L. M., MacKenzie, E., Vega, I. P., Vanzella, L., Fjeldstad, J., Acera, A. M., Cabezuelo, V., Bueno, B., Rexach, A., & Bermejo, A. (2026). L26/P-229 Artificial intelligence (AI)-derived oocyte ploidy score predicts PGT-A outcome beyond conventional morphology, with increasing value at advanced maternal age. Human Reproduction, 41(Supplement_1). https://doi.org/10.1093/humrep/deag083.565
Vancouver
Iriarte YF, Villa GB, Milla AV, Burgos LB, Borras FS, Aguilar LM, et al. L26/P-229 Artificial intelligence (AI)-derived oocyte ploidy score predicts PGT-A outcome beyond conventional morphology, with increasing value at advanced maternal age. Human Reproduction. 2026;41(Supplement_1). doi:10.1093/humrep/deag083.565
BibTeX
@article{Iriarte2026,
author = {Iriarte, Y. Franco and Villa, G. Bescós and Milla, A. Villa and Burgos, L. Barroso and Borras, F. Sotos and Aguilar, L. Murria and MacKenzie, E. and Vega, I. Puerta and Vanzella, L. and Fjeldstad, J. and Acera, A. Marinez and Cabezuelo, V. and Bueno, B. and Rexach, A. and Bermejo, A.},
title = {L26/P-229 Artificial intelligence (AI)-derived oocyte ploidy score predicts PGT-A outcome beyond conventional morphology, with increasing value at advanced maternal age},
journal = {Human Reproduction},
year = {2026},
volume = {41},
number = {Supplement_1},
doi = {10.1093/humrep/deag083.565}
}
EUVIMED