A data-driven approach for cell culture medium optimization
Yuki Ozawa, Takamasa Hashizume, Bei‐Wen Ying
Abstract
Cell culture media are critical for cell propagation and bioproduction to be as efficient as possible to meet medical and pharmaceutical requirements. However, optimizing medium composition to achieve optimal cell culture remains a significant challenge due to the complexity of living cells and their required media. This study, which is a significant contribution to the field, addresses the need for data-driven techniques in cell culture technologies by integrating active machine learning (ML) to reformulate a widely used base medium for mammalian cell culture. The optimization process was facilitated by developing various ML models, which accounted for experimental data processing and time consumption. It provided a detailed methodology and essential knowledge for utilizing ML in medium development. Growth determinative medium components were identified through data mining and scale-up culture. In addition, RNA sequencing analysis indicated that active learning finetuned the media for the changes in gene expression for improved cell culture. This study offers new insights and methodologies to be applied to cell culture for future medical purposes. • This study represents a significant step forward in applying machine learning to optimizing mammalian cell culture and tailored transcriptome response. • This data-driven optimization approach not only enhances medium formulation for bioproduction but also provides a robust methodology for future medical applications, potentially revolutionizing the way we approach cell culture in personalized medicine.