Limitations of a time-lapse blastocyst prediction model: a large multicentre outcome analysis

The goal of embryo selection models is to select embryos with the highest reproductive potential, whilst minimizing the rejection of viable embryos. Ultimately, any embryo selection model must be tested on clinical outcome. We therefore retrospectively tested a published blastocyst prediction model on a large combined set of transferred embryos with known clinical outcome. The model was somewhat effective in that we found a relative increase of 30% for implantation in the model-selected group of embryos. There was, however, a concomitant large rejection of embryos from our test cohort, which actually resulted in pregnancy. This hypothetical experiment highlights the limitations of predicting blastulation only. Crucially, it illustrates that both sensitivity and specificity are important parameters when developing embryo selection models for prospective clinical use.