Multifactor Prediction of Embryo Transfer Outcomes Based on a Machine Learning Algorithm.

Area of Science:

• Reproductive Medicine
• Clinical Data Science
• Biostatistics

Background:

• In vitro fertilization-embryo transfer (IVF-ET) is a key technology for infertile couples, but success is not guaranteed.
• Frozen embryo transfer (FET) is a crucial component of IVF-ET, yet its outcomes remain unpredictable.
• Machine learning (ML) offers potential for predicting clinical outcomes by analyzing complex datasets.

Purpose of the Study:

• To develop and evaluate machine learning models for predicting early pregnancy outcomes in FET.
• To identify key clinical factors influencing FET success.
• To provide a predictive reference for couples undergoing FET.

Main Methods:

• Utilized clinical data including age, BMI, endometrial thickness (EMT), good-quality embryo rate (GQR), infertility type, serum estradiol (E2), and progesterone (P).
• Applied four ML algorithms: logistic regression (LR), conditional inference tree, random forest (RF), and support vector machine (SVM).
• Assessed model performance using sensitivity, specificity, positive predictive rate, and negative predictive rate.

Main Results:

• No significant differences were observed among the four ML models in predicting FET pregnancy outcomes.
• The Support Vector Machine (SVM) model demonstrated a positive predictive rate of 0.56 and a negative predictive rate of 0.55.
• The overall prediction accuracy was limited, indicating potential for further predictor discovery.

Conclusions:

• Current ML models using selected clinical variables have limited accuracy in predicting FET pregnancy outcomes.
• The study highlights the need for identifying novel and more effective predictive factors for FET success.
• This research provides a foundational reference for couples considering FET, while emphasizing areas for future investigation.