Improving embryo selection by the development of a laboratory-adapted time-lapse model

Area of Science:

• Reproductive Medicine
• Embryology
• In Vitro Fertilization (IVF)

Background:

• General embryo selection models are widely used in IVF laboratories.
• These models may not fully capture laboratory-specific embryo development patterns.
• Optimizing embryo selection is crucial for improving IVF success rates.

Purpose of the Study:

• To develop and validate a powerful, in-house embryo selection model for a specific IVF laboratory.
• To assess if a laboratory-specific model can improve upon general embryo selection models.
• To analyze the contribution of specific morphokinetic parameters to embryo selection.

Main Methods:

• Analysis of 12,944 fertilized oocytes and 1,879 embryos with known implantation data (KID).
• Comparison of cell division timings between KID-positive and -negative embryos for IVF and intracytoplasmic sperm injection (ICSI).
• Development of embryo selection algorithms for days 2, 3, and 5 using EmbryoScope Stats software and validation via receiver operating characteristic (ROC) curve analysis (AUC).

Main Results:

• Significant differences in morphokinetics were identified between KID-positive and -negative embryos.
• In-house models developed for ICSI embryos on days 2, 3, and 5 showed validity with AUC values generally above 0.65.
• The day 5 model demonstrated the highest predictability with an AUC of 0.803, indicating improved performance over days 2 and 3.

Conclusions:

• A laboratory-specific morphokinetic model is a valuable addition to general models for enhancing single embryo selection in IVF.
• The development process provided insights that can facilitate similar model development in other laboratories.
• In-house models can improve IVF outcomes by better characterizing embryo potential within a specific laboratory context.