Prediction of Cardiac Differentiation in Human Induced Pluripotent Stem Cell-Derived Cardiomyocyte Supernatant using Surface-Enhanced Raman Spectroscopy and Machine Learning

The efficient manufacturing of cardiomyocytes from human-induced pluripotent stem cells (hiPSCs) is essential for advancing regenerative therapies for myocardial injuries. However, ensuring cell quality during production is challenging since traditional methods are invasive, destructive, and time-consuming. In this study, we monitored cardiomyocyte differentiation of WTC11 hiPSCs by analyzing conditioned media collected at various stages using Raman spectroscopy, multivariate analysis, and machine learning. Differentiation efficiency was confirmed via flow cytometry and immunostaining. Raman spectra were processed using standard normal variate and second derivative transformations before performing a principal component analysis (PCA) and machine learning (Random Forest, K-Nearest Neighbors, and Deep Neural Networks [DNN]). Results show that PCA was unable to distinguish cells based on differentiation stages, while machine learning could reliably predict cell differentiation early in the cardiac cell manufacturing process. DNN models achieved accuracies exceeding 82 % in predicting differentiation, highlighting their potential as quality control tools. These findings underscore the potential of Raman spectroscopy coupled with machine learning as a tool for real-time monitoring of cardiomyocyte production.