부산대학교 도서관

This article explores the use of machine learning (ML) models in predicting the ploidy status of embryos in in vitro fertilization (IVF). The study compares different ML models using morphokinetic and clinical data to predict the ploidy status of blastocysts. The authors find that a traditional logistic regression model performs better than more complex ML algorithms when predicting aneuploidy, but its performance decreases when predicting euploidy. The article also discusses the impact of dataset characteristics, the relevance of clinical variables, potential bias in ML models based on oocyte age, and the influence of outcome imbalance on model predictions. The authors caution that ML models should be used cautiously and in conjunction with other clinical factors, and emphasize the importance of interdisciplinary collaboration and ethical considerations in their use. [Extracted from the article]