Joint Research on Early Non-Destructive Prediction of Human iPSC Differentiation Efficiency Using Image Analysis and Machine Learning Published in Scientific Reports

Epistra Inc. (Tokyo) has jointly developed a novel method with the Center for iPS Cell Research and Application (CiRA) at Kyoto University that combines cell image analysis and machine learning. This method enables early, non-destructive prediction of differentiation efficiency from human iPSCs to muscle stem cells (MuSCs). The research findings were published online in Scientific Reports on July 23, 2025.

Overview and Results

Human iPSCs have the ability to differentiate into various cell types and hold promise for applications in regenerative medicine and drug discovery. The “stepwise differentiation induction method” is advantageous in terms of safety due to its high maturity, but variability in efficiency and extended culture periods (several months) remain challenges.

In this joint research, MuSC differentiation induction was used as a model system. By combining image analysis and machine learning, a method was developed to predict final differentiation efficiency early and non-destructively. While MuSC differentiation typically requires approximately 80 days, this method can predict efficiency at days 24-34, potentially contributing to efficient cell acquisition and accelerating regenerative medicine research.

Key advantages include avoiding destructive analytical methods, eliminating subjective expert evaluation, and enabling rapid and objective determination of differentiation status using images alone.

Paper Information

• Journal: Scientific Reports
• DOI: 10.1038/s41598-025-11108-5