L26/P-275 Comprehensive assessment of a deep learning tool for the prediction of embryo ploidy status
Human Reproduction · 01.07.2026 · 41 (Supplement_1)
DOI: 10.1093/humrep/deag083.611 ↗Kurz gefasst
Fachartikel in der Zeitschrift Human Reproduction, online erschienen am 08.07.2026. Verfasst von I. Sfontouris und 6 weiteren Autor:innen.
Abstract (Originalsprache)
Study question Can the deep learning tool iDAScore v2 predict blastocyst ploidy status? Summary answer iDAScore alone could discriminate euploid from aneuploid blastocysts with an AUC of 0.684, and a model combining iDAScore and Age increased the AUC to 0.763. What is known already iDAScore is an automated decision-support system that uses deep learning to analyze time-lapse embryo images and rank embryos according to their implantation potential. Notably, iDAScore was also shown to correlate with embryo ploidy status. Earlier versions of iDAScore (v1) demonstrated a moderate capacity to predict embryo aneuploidy. This study thoroughly provides a comprehensive assessment of iDAScore version 2 for the prediction of embryo ploidy status. Study design, size, duration Retrospective study of 1237 blastocysts of known ploidy from 326 patients undergoing PGT-A. For statistical analysis, continuous data were compared between euploidy categories using Dunnett’s test. Receiver operating characteristic (ROC) curve analysis was used to estimate the discrimination performance of variables to predict aneuploidy. Logistic regression was used to construct models for aneuploidy prediction based on age and iDAScore. Participants/materials, setting, methods Patients eligible for PGT-A had either 2 or more previous failed transfers, 2 or more pregnancy losses, or were aged 40 years and above, according to national legislation. Median female age was 40.7 years. Embryos were cultured in an Embryoscope+ and biopsied on Day5/6 and then vitrified. Whole-genome amplification and next-generation sequencing were used for PGT-A. iDAScore was calculated using version 2.0. Main results and the role of chance Out of 1237 embryos analysed, 30.1% were euploid (EUPL), 52.1% single aneuploid (S-ANEUP), 10.0% complex aneuploid (C-ANEUP), 7.9% mosaic (MOS). The IDAScore median (Q25%-Q75%) for EUPL was 7.1 (4.4-8.4); S-ANEUP 4.7 (2.6-7.1); C-ANEUP 3.3 (1.5-5.3); MOS 5.7 (2.9-8.0). The EUPL median was significantly higher (p < 0.01) compared to all other categories. Embryo quality declined with age which was reflected by a decline in iDAScore (<35 years 6.7 (3.5-8.1); 35-39 years 5.8 (3.1-8.0); 40-42 years 5.0 (2.9-7.6); >42 years 4.6 (2.2-7.1). A model based on female age only discriminated between euploids and non-euploids with an AUC [95% C.I.] of 0.704 [0.671-0.737]. For iDAScore alone the discrimination had an AUC of 0.684 [0.646-0.721]. When combining iDAScore and female age the performance increased to 0.763 [0.730-0.796]. When classifying cases by day of biopsy, the AUC of iDAScore to identify euploid embryos was 0.709 [0.674 to 0.743] for Day 5 and 0.671 [0.573 to 0.769] for Day 6. The combined iDAScore and female age model yielded increased AUC for euploidy prediction 0.776 [0.742 to 0.810] for Day 5 and 0.696 [0.605 to 0.787] for Day 6. Limitations, reasons for caution This is a retrospective single center study and therefore the presence of selection bias cannot be excluded. The estimated models can be used as screening tools but do not replace PGT-A biopsies. Wider implications of the findings iDAScore demonstrates a moderate yet significant capacity to predict embryo aneuploidy. It may inform decisions regarding transfer and biopsy ranking, and, when combined with age, may be used during patient consultation to discuss the overall risk of aneuploidy. Trial registration number No
Zitieren
APA
Sfontouris, I., Papadogianni, P., Tsorva, E., Berntsen, J., Johansen, M., Kerr, R., & Makrakis, E. (2026). L26/P-275 Comprehensive assessment of a deep learning tool for the prediction of embryo ploidy status. Human Reproduction, 41(Supplement_1). https://doi.org/10.1093/humrep/deag083.611
Vancouver
Sfontouris I, Papadogianni P, Tsorva E, Berntsen J, Johansen M, Kerr R, et al. L26/P-275 Comprehensive assessment of a deep learning tool for the prediction of embryo ploidy status. Human Reproduction. 2026;41(Supplement_1). doi:10.1093/humrep/deag083.611
BibTeX
@article{Sfontouris2026,
author = {Sfontouris, I. and Papadogianni, P. and Tsorva, E. and Berntsen, J. and Johansen, M. and Kerr, R. and Makrakis, E.},
title = {L26/P-275 Comprehensive assessment of a deep learning tool for the prediction of embryo ploidy status},
journal = {Human Reproduction},
year = {2026},
volume = {41},
number = {Supplement_1},
doi = {10.1093/humrep/deag083.611}
}
EUVIMED