Predicting Usable Blastocyst Using Morphological and Morphokinetic Parameters on Day 2 Embryo Development through Machine Learning

Abstract

Background and Aims: Efficient prediction of blastocyst formation from early-stage human embryos is valuable. In this context, machine learning (ML) methods were developed to assist embryologists with automatized and objective predictive models able to standardize human embryo assessment. In this study, we aimed to develop and validate an ML model to predict usable blastocysts based on morphological and morphokinetic Day 2 parameters. Methods: This retrospective study analyzed the morphokinetics of 1,910 embryos obtained from 232 couples who underwent ICSI. Embryo morphokinetics were recorded using the Astec™time-lapse system. A total of 18 variables, including clinical, morphological, and morphokinetic, from the first two days of embryo development were recorded and combined. An Extreme Gradient Boosting (XGBoost) model was trained and validated with model performance evaluated using accuracy. Feature importance was analyzed using Shapley Additive Explanations (SHAP) values. Results: The final XGBoost accurately predicted blastocyst usability, achieving an area under the curve (AUC) of 0.77 and an accuracy of 71%. The most significant predictive variables included: The time of the second cell cycle (cc2), the time of division to two cells (t2), blastomere number, fragmentation rate, and maternal age. Conclusions: The XGBoost ML model uses the morphological and morphokinetic features of early embryonic development to predict usable blastocysts effectively with an accuracy of 71%. This study demonstrates the potential of ML models to change embryo selection in ART.