Deep Learning to Predicting Live Births and Aneuploid Miscarriages from Images of Blastocysts Combined with Maternal Age

Abstract

Objectives: Making an artificial intelligence (AI) classifier that uses the maternal age and an image of the implanted blastocyst to determine the probability of getting a live birth. Methods: The dataset comprised maternal age data and 407 images of blastocysts which led to live births and 246 images of blastocysts which led to aneuploid miscarriages, matched for maternal age. An AI system using deep learning was developed for predicting the classification and probability of a live birth. Results: The accuracy, sensitivity, specificity, and positive and negative predictive values of the developed AI classifier were 0.75, 0.82, 0.64, 0.79, and 0.68, respectively. The area under the curve was 0.73 ± 0.04 (mean ± standard error). Conclusions: A classifier using AI for a blastocyst image combined with the maternal age showed potential in determining the probability of a live birth.