Life Whisperer AI blastocyst assessment predicted implantation across 340 frozen embryo transfer cycles with AUC 0.91, but concordance with PGT-A was only κ=0.257

AI-assisted blastocyst viability scoring independently predicted implantation outcomes in frozen embryo transfer cycles, with implantation rates rising from 53.5% in the low-viability group to 74.2% in the high-viability group and an area under the ROC curve of 0.91. Whereas embryo assessment has relied primarily on Gardner morphological grading, this study evaluated Life Whisperer™ AI viability scoring alongside Gardner grading and reported its independent predictive performance across 340 cycles. Based on a retrospective cohort of 340 frozen embryo transfer cycles, with implantation determined by serum β-hCG positivity and discriminatory performance assessed by ROC analysis, reporting an AUC of 0.91.

Higher Gardner morphological grades were associated with superior implantation rates, with grades 4AA and 3AB showing implantation rates of 80.7% and 81.0%, respectively. The study presents conventional morphological grading and AI scoring within the same cohort, providing comparative data for both approaches in the frozen embryo transfer setting. Derived from retrospective data on 340 cycles, with implantation outcomes determined by serum β-hCG positivity.

AI-derived genetic prediction showed only fair concordance with PGT-A results, with Cohen's κ = 0.257. The study directly tested concordance between AI genetic prediction and PGT-A rather than only assessing implantation prediction, thereby clarifying the positioning of AI in genetic screening. Concordance was assessed using Cohen's kappa coefficient, reporting κ = 0.257, indicating a fair level of agreement.

The study defines the setting in which AI blastocyst assessment applies: as a complement to conventional morphological evaluation for optimizing embryo selection and improving implantation outcomes in frozen embryo transfer cycles. Beneficiaries may include embryologists and clinicians in assisted reproduction who need more objective, non-invasive embryo assessment, under conditions of Day 5 blastocysts, frozen embryo transfer cycles, and implantation determined by serum β-hCG positivity. The study also states that AI-derived genetic prediction cannot currently substitute for PGT-A in genetic screening.

Readers should still watch: this is a single-center retrospective cohort of 340 cycles, and whether results generalize to other centers, populations, or transfer protocols remains to be validated; AI-derived genetic prediction showed only fair concordance with PGT-A (κ = 0.257), and its role in genetic screening requires further clarification; additionally, the loaded text is an incomplete version lacking figures and supplementary materials, so finer stratified data, confidence intervals, or subgroup analyses could not be verified and await confirmation in the original article.