A multiday machine learning framework based on improved-NEQsi for predicting embryo quality and pregnancy outcomes in...

Area of Science:

• Reproductive Medicine
• Embryology
• Artificial Intelligence in Healthcare

Background:

• Current embryo assessment methods are fragmented and resource-intensive.
• Integrating clinical factors with embryo morphology is crucial for improved IVF outcomes.
• Limitations exist in current AI-based embryo selection and clinical data integration.

Purpose of the Study:

• To develop an Improved Numerical Embryo Quality scoring index (INEQsi) for multi-day embryo assessment.
• To create a lightweight machine learning tool for predicting embryo quality and pregnancy success.
• To integrate standardized embryo scores with clinical data for enhanced IVF decision-making.

Main Methods:

• A multidimensional embryo scoring system (INEQsi) was developed based on established grading standards.
• Clinical variables were integrated with INEQsi scores to build a machine learning prediction model.
• Multiple algorithms were evaluated to determine the optimal model for predicting embryo quality and pregnancy outcomes.

Main Results:

• The random forest model using INEQsi demonstrated high accuracy in predicting embryo quality (RMSE: 1.11-1.26) across cleavage and blastocyst stages.
• INEQsi outperformed NEQsi in predicting pregnancy outcomes, achieving 99-100% sensitivity on Day 5 and 93% accuracy on Day 6.
• The model showed predictive capability extending to Day 3 embryo assessment.

Conclusions:

• A multitage embryo scoring system (INEQsi) combined with machine learning offers a lightweight analytical framework for IVF.
• This approach shows potential as a supplementary tool for embryo assessment and clinical pregnancy prediction.
• The framework is suitable for settings where advanced imaging systems are not routinely available.