Can artificial intelligence guided feedback improve embryologists' selection of euploid embryos based on morphology...

Area of Science:

• Reproductive Medicine
• Embryology
• Artificial Intelligence in Healthcare

Background:

• Embryo selection is crucial for successful in vitro fertilization (IVF).
• Current methods rely heavily on embryologist expertise in embryo morphology assessment.
• Subjectivity and inter-observer variability can impact embryo selection accuracy.

Purpose of the Study:

• To evaluate the reliability of embryologists in differentiating euploid from aneuploid embryos based on morphology alone.
• To assess if artificial intelligence (AI)-assisted selection using ERICA (Embryo Ranking Intelligent Classification Algorithm) improves embryo selection outcomes compared to embryologists alone.
• To investigate the impact of AI on human decision-making in IVF embryo selection.

Main Methods:

• 19 embryologists with varying experience levels evaluated embryo images for ploidy status.
• Embryologists were provided with AI-generated rankings from ERICA for the same embryos.
• A training tool was used, and embryologists made final judgments combining their assessment with AI rankings across simulated IVF cycles.

Main Results:

• Both embryologists and ERICA demonstrated statistically significant ability to identify aneuploidy above random chance.
• ERICA showed superior performance in selecting euploid embryos on the first attempt compared to embryologists (P=0.002).
• No significant difference was found in embryologists' ability to select better embryos, and AI assistance did not significantly alter their final decisions.

Conclusions:

• AI tools like ERICA can enhance embryo selection reliability by reducing subjectivity and bias in IVF.
• The combination of human and AI judgment does not consistently offer an advantage over independent use.
• Further research is needed to understand AI's influence on human decision-making and trust in automated IVF processes.