Related Experiment Video
Updated: Nov 10, 2025
05:13
Author Spotlight: Advancing Therapeutic Strategies for Improving Pregnancy Rates by Analyzing Embryo-Endometrium Interactions
Published on: June 21, 2024
1.3K
Individualized embryo selection strategy developed by stacking machine learning model for better in vitro
Qingsong Xi1, Qiyu Yang1, Meng Wang1
1Reproductive Medicine Center, Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology, No.1095, Jiefang Road, Wuhan, 430030, China.
Reproductive Biology and Endocrinology : RB&E
|April 6, 2021
Summary
This study introduces an AI model to optimize embryo selection in in vitro fertilization (IVF), predicting pregnancy and twin risks for personalized treatment. The XGBoost model aids in deciding between single embryo transfer (SET) and double embryo transfer (DET).
Area of Science:
- Reproductive Medicine and Assisted Reproductive Technologies
- Artificial Intelligence in Healthcare
- Embryology and In Vitro Fertilization
Background:
- Minimizing multiple-embryo gestation in in vitro fertilization (IVF) is crucial.
- Selecting between single embryo transfer (SET) and double embryo transfer (DET) is challenging, especially for patients with sub-optimal prognosis or lower-quality embryos.
- Existing machine learning (ML) applications in IVF primarily focus on selecting top-quality embryos.
Purpose of the Study:
- To develop and evaluate an AI-based hierarchical model using XGBoost for embryo implantation potential.
- To simultaneously assess the impact of double embryo transfer (DET) on IVF outcomes.
- To provide an individualized embryo selection strategy to optimize clinical pregnancy rates and minimize twin risks.
Main Methods:
- An application study involving 9,211 patients and 10,076 embryos treated between 2016 and 2018.
- A hierarchical XGBoost model was established to predict embryo implantation potential and DET impact.
- Model performance was evaluated using Area Under the Curve (AUC) of the Receiver Operating Characteristic (ROC) curve; multiple regression analyses identified key predictive features.
Main Results:
- Significant predictors for SET pregnancy included age, IVF attempts, estradiol level, and endometrial thickness.
- For DET pregnancy, age, IVF attempts, endometrial thickness, and P1+P2 were significant.
- Significant predictors for DET twin risk were age, IVF attempts, 2PN/MII, and P1×P2. The XGBoost model achieved AUCs of 0.7945 (SET pregnancy), 0.8385 (DET pregnancy), and 0.7229 (DET twin risk), outperforming logistic regression and classification and regression trees.
Conclusions:
- AI-based determinant-weighting analysis can create individualized embryo selection strategies.
- The model accurately predicts clinical pregnancy rates and twin risks, optimizing IVF outcomes.
- This approach supports informed decisions regarding SET versus DET, enhancing patient care.
Keywords:
Artificial intelligenceEmbryo selectionIn vitro fertilizationIn vitro fertilization predictionMachine learning
