Artificial intelligence-based donor oocyte quality assessment moderately improves the prediction of blastocyst development: a first step towards higher personalization in the management of egg donation treatments
Bibliographic record
Abstract
STUDY QUESTION: Can an artificial intelligence (AI)-based oocyte scoring system reliably predict the developmental competence of fresh donor oocytes? SUMMARY ANSWER: The AI-derived Magenta Score was significantly associated with fertilization, blastocyst formation, and helpful to estimate cumulative live birth rates, although a trend toward overestimation was observed in a subset of cycles. WHAT IS KNOWN ALREADY: Oocyte quality is a critical determinant of IVF success; however, standardized and objective methods for its assessment are lacking. Current allocation strategies in oocyte donation cycles often neglect recipient-related factors and risk overproduction of surplus embryos. AI-based evaluation may offer a more objective, reproducible alternative to traditional morphology-based assessment. STUDY DESIGN, SIZE, DURATION: Prospective, observational, multicenter, blinded cohort study including 1179 fresh metaphase II (MII) oocytes from 145 donors, allocated to 171 recipient couples across three IVF centers between June 2023 and October 2024. PARTICIPANTS/MATERIALS, SETTING, METHODS: Denuded MII oocytes were imaged at 200-400× magnification and assessed using an AI-based scoring system (Magenta Score, Future Fertility). The primary outcome was the association between Magenta Score and blastocyst development, adjusted for donor age, sperm motility, and culture medium. Secondary outcomes included associations with oocyte dysmorphisms, fertilization, blastocyst quality and timing, implantation, cumulative live birth rates, and accuracy of blastocyst yield predictions. MAIN RESULTS AND THE ROLE OF CHANCE: Oocytes with higher Magenta Scores had significantly higher rates of 2PN fertilization (odds ratio [OR] 1.08) and blastocyst formation (OR 1.19), independent of confounders. Magenta Score per se displayed an AUC of 0.6, reaching 0.62 if combined with donors' age and 0.65 if also combined with male partners' sperm motility 1%-increase and culture medium used, highlighting the multifactorial nature of embryo development. In 82% of cases, the actual blastocyst number fell within or above the predicted range extrapolated from the Magenta Scores of each cohort. A 10% increase in the predicted probability of achieving at least one live birth based on the Magenta Score was associated with a significantly higher true cumulative live birth rate (OR 1.55; AUC 0.691). LIMITATIONS, REASONS FOR CAUTION: The observational design precludes causal inference. Only fresh oocyte cycles were evaluated, limiting extrapolation to vitrified oocytes. Some donor oocytes were cryopreserved and excluded from analysis. Future randomized trials are needed to assess clinical utility when AI is actively used for allocation decisions. WIDER IMPLICATIONS OF THE FINDINGS: AI-based assessment of donor oocytes offers a promising tool to enhance the personalization and fairness of oocyte allocation in donation cycles. However, to maximize its clinical value, AI predictions should be integrated with additional donor-, recipient-, and cycle-specific variables. Further refinements and prospective validations are necessary to improve prediction accuracy and avoid overestimation, ultimately optimizing cumulative live birth rates while minimizing surplus embryo production. STUDY FUNDING/COMPETING INTEREST(S): No funding. N.M., J.F., D.N., and A.K. are employees and hold stock options of Future Fertility, the company that developed the AI model used. All other authors report no conflict of interest related with the content of this manuscript. TRIAL REGISTRATION NUMBER: n/a.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.002 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one teacher head, not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".