Using deep learning to predict the outcome of live birth from more than 10,000 embryo data

Abstract

Background Recently, the combination of deep learning and time-lapse imaging provides an objective, standard and scientific solution for embryo selection. However, the reported studies were based on blastocyst formation or clinical pregnancy as the end point. To the best of our knowledge, there is no predictive model that uses the outcome of live birth as the predictive end point. Can a deep learning model predict the probability of live birth from time-lapse system? Methods This study retrospectively analyzed the time-lapse data and live birth outcomes of embryos samples from January 2018 to November 2019. We used the SGD optimizer with an initial learning rate of 0.025 and cosine learning rate reduction strategy. The network is randomly initialized and trained for 200 epochs from scratch. The model is quantitively evaluated over a hold-out test and a 5-fold cross-validation by the average area under the curve (AUC) of the receiver operating characteristic (ROC) curve. Results The deep learning model was able to predict live birth outcomes from time-lapse images with an AUC of 0.968 in 5-fold stratified cross-validation. Conclusions This research reported a deep learning model that predicts the live birth outcome of a single blastocyst transfer. This efficient model for predicting the outcome of live births can automatically analyze the time-lapse images of the patient’s embryos without the need for manual embryo annotation and evaluation, and then give a live birth prediction score for each embryo, and sort the embryos by the predicted value.

Prediction models and associated factors on the fertility behaviors of the floating population in China

“…A study on willingness of the floating population to have a second child in Hunan Province found that the relevant factors of fertility willingness included gender, age, occupation, education level, and marital status ( 15 ). Logistic regression, neural networks, and other machine learning models had been used to predict the birth results of pregnant women ( 16 ) and live birth results of embryos ( 17 ). However, there was still a lack of model research used to predict the fertility behavior of the floating population.…”

Measuring cumulus expansion of mammalian oocytes: comparing the reliability of methods and how artificial intelligence could automate the measurement

“…In recent years, assisted reproductive technologies have bene ted from the increased use of arti cial intelligence techniques such as DL and image segmentation 34,[48][49][50][51][52][53] , although segmenting images of bovine COCs had not been done yet. AIxpansion was therefore developed to automatically measure the area of immature and mature bovine COCs by image segmentation, and to consequently calculate the relative expansion of the cumulus cells.…”

Delineating the heterogeneity of preimplantation development via unsupervised clustering of embryo candidates for transfer using automated, accurate and standardized morphokinetic annotation

“…[23][24][25] Parallel efforts took advantage of the size of the available time-lapse datasets to train convolutional neural network (CNN) based classifiers that assess embryo potential using the raw video files in an unbiased annotation-independent manner. 26,27 However, training such deep learning models is challenging due to the size of the video files (~100's Mb), which would require a sufficiently large sample number. 28,29 Morphokinetic evaluation of the developmental potential proved highly efficient in de-selecting for transfer poor quality embryos.…”