Predicting pregnancy outcomes in IVF cycles: a systematic review and diagnostic meta- analysis of artificial intelligence in embryo assessment.

Abstract

Introduction: Embryo selection remains a key challenge in in vitro fertilization (IVF), as many morphologically "normal" embryos fail to implant. Artificial intelligence (AI) offers a promising tool for improving embryo assessment by providing more objective and accurate predictions of pregnancy outcomes. This study aims to systematically review and conduct a diagnostic meta-analysis to evaluate the effectiveness of AI-based tools in embryo selection for predicting pregnancy outcomes in IVF.

Methods: We conducted a systematic review following PRISMA guidelines, searching Web of Science, Scopus, and PubMed. Original research articles evaluating AI's diagnostic accuracy in embryo selection were included, while duplicates, non-peer-reviewed papers, abstracts, and conference proceedings were excluded. Data on sample sizes, AI tools, and diagnostic metrics were extracted, with quality assessed using the QUADAS-2 tool.

Results: AI-based embryo selection methods showed strong diagnostic performance, with pooled sensitivity of 0.69 and specificity of 0.62 in predicting implantation success. The positive likelihood ratio was 1.84 and the negative likelihood ratio was 0.5. The area under the curve reached 0.7, indicating high overall accuracy. The Life Whisperer AI model achieved 64.3% accuracy in predicting clinical pregnancy, while the FiTTE system, which integrates blastocyst images with clinical data, improved prediction accuracy to 65.2% with an AUC of 0.7.

Conclusion: AI offers a promising advancement in embryo selection for IVF, with the potential to enhance clinical outcomes and improve decision-making. Future studies should focus on refining these models to achieve the ultimate goal of a healthy live birth by developing more sophisticated algorithms and validating them with larger, diverse datasets.