JBRA Assisted Reproduction 2025;29(Suppl.2 SBRA 2025):5
Oral Presentation
29th Annual Congress of the SBRA. São Paulo/SP - Brazil, 2025
doi: 10.5935/1518-0557.20263519
O-05. Evaluating AI-Based Oocyte and Embryo Scoring as a Non-Invasive Approach for Ploidy Assessment in ART
Mariângela Badalotti1, Victoria Dornelles2, Fabiana Wingert1, Marta Ribeiro Hentschke1, Ricardo Azambuja1, Isadora Badalotti-Teloken1, Carolina Comissoli Fernandes2, Bárbara dal Magro1, Laura Chapochnicoff2, Júlia Prauchner Castilhos2, Álvaro Petracco1
1 Fertilitat - Centro de Medicina Reprodutiva - Porto Alegre - RS - Brasil
2 Pontificia Universidade Católica do Rio Grande do Sul - Porto Alegre - RS - Brasil
Objective: To evaluate whether AI-based oocyte and embryo scoring can serve as a non-invasive predictor of embryo ploidy in assisted reproductive technologies (ART).
Methods: This retrospective case-control study included 347 biopsied embryos from ART patients treated between January and September 2024 at a reproductive center in Brazil. Embryos were categorized by PGT results into G1 (euploid, n=107) and G2 (aneuploid/mosaic, n=240). Patients with maternal age >38 years were excluded to reduce confounding effects. Oocytes were scored using Magenta IVF®, and blastocysts were evaluated using KIDScore™. Statistical analyses included Student’s t-test, Mann–Whitney U test, Spearman correlation, binary logistic regression, and ROC curve analysis (significance level set at p<0.05).
Results: No significant differences were found between G1(euploid) and G2 (aneuploid/mosaic) in oocyte age (32.8±4.52 vs. 34.10±3.52; p=0.079), sperm age (35.0±2.70 vs. 36.8±5.18; p=0.085), or sperm concentration (59.81±52.09 vs. 57.31±33.48 million/mL; p=0.899). Mean Magenta scores did not differ significantly between groups (6.63±2.43 vs. 6.50±2.45; p=0.707), whereas mean KIDScore™ was significantly higher in the euploid group (5.58±1.92 vs. 4.69±1.80; p<0.001). Binary logistic regression adjusted for patient age showed that higher KIDScore™ values were associated with lower odds of aneuploidy (OR=0.754; 95% CI: 0.61–0.91; p=0.005). This association was not observed for Magenta scores (OR=1.019; 95% CI: 0.87–1.18; p=0.804). ROC analysis confirmed KIDScore™ as a modest predictor of euploidy (AUC=0.633; p<0.001), while Magenta scoring showed no predictive value (AUC=0.513; p=0.707). No correlation was found between Magenta and KIDScore™ in either group (G1: r=–0.21; p=0.832; G2: r=0.63; p=0.333).
Conclusion: While AI-based oocyte scoring (Magenta IVF®) did not correlate with ploidy, KIDScore™ was significantly associated with euploidy and embryo quality. These findings suggest a potential role for AI-based embryo evaluation in supporting clinical decision-making. However, current AI tools are not yet substitutes for PGT. Further studies with larger cohorts and more advanced algorithms are warranted to strengthen the predictive value of non-invasive embryo assessment in ART.