Zachary A. Cosenza (University of California): Design of Cell Culture Media with Multi-Information Source Bayesian Optimization

Graduate Student - 03, LIKE22: videos of early stage researchers

Culture media used in industrial bioprocessing is difficult to optimize due to the lack of rigorous mathematical models of the relationship between cell growth and culture conditions as well as the size of the relevant design space. In this work, we optimized a 14-dimensional cell growth culture media using a multi-information source Bayesian optimization algorithm that locates optimal media conditions based on an iterative refinement of an uncertainty weighted desirability function. Using this algorithm, we were able to design a media 2.8x better than a common commercial variant at nearly the same cost, while doing so in 38% fewer experiments than a sophistical design-of-experiments method. Due to the choice of a robust multi-assay objective function, these results generalized well to long term multi-passage cell growth.