Predictive models for live birth outcomes following fresh embryo transfer in assisted reproductive technologies using machine learning.

Abstract

Background: Infertility affects approximately 15% of couples globally, with assisted reproductive technologies (ARTs) becoming the primary interventions. Despite the growing use of ARTs, success rates have plateaued at around 30%, highlighting the need for improved predictive models to enhance outcomes. This study aimed to develop a machine learning-based predictive model for live birth outcomes following fresh embryo transfer.

Methods: A total of 51,047 ART records were collected from 2016 to 2023 at the Shanghai First Maternity and Infant Hospital. After data preprocessing, 11,728 records and 55 pre-pregnancy features were analyzed. Six machine learning models-Random Forest (RF), eXtreme Gradient Boosting (XGBoost), Gradient Boosting Machines (GBM), Adaptive Boosting (AdaBoost), Light Gradient Boosting Machine (LightGBM), and Artificial Neural Network (ANN)-were employed to construct the prediction model.

Results: Among the models, RF demonstrated the best predictive performance, achieving an area under the curve (AUC) value exceeding 0.8. Key predictive features included female age, grades of transferred embryos, number of usable embryos, and endometrial thickness. A web tool was developed to assist clinicians in predicting outcomes and individualizing treatments based on patient data.

Conclusions: This study presents a significant advancement in predicting live birth outcomes prior to embryo transfer, moving beyond traditional assessments. The findings underscore the potential of machine learning to improve clinical decision-making and enhance patient counseling in ARTs.