Aplicación de la inteligencia artificial en la selección embrionaria durante la fertilización in vitro

Embryo selection in in vitro fertilization relies on a subjective morphological assessment that varies between observers. The study therefore aimed to systematise the evidence on artificial intelligence in embryo selection during in vitro fertilization, its clinical efficacy, diagnostic performance and reproducibility. A PRISMA 2020-guided systematic review was conducted in PubMed and SciELO, with a sample of 18 studies out of 187 records. The iDAScore v2.0 system achieved a clinical pregnancy rate of 46.5% versus 48.2% for morphological assessment in a randomised trial, without formal non-inferiority; artificial intelligence predicted pregnancy with an accuracy of 77.8% with images and 81.5% with clinical data, versus 64% and 51% for embryologists. It was concluded that trials with live birth as the outcome are needed.