- Cite This Publication
Who: Zachary Cosenza, Raul Astudillo, Peter I. Frazier, Keith Baar, David E. Block
Published: May 11, 2022
Where: Biotechnology and Bioengineering
Key Takeaway: Bayesian optimization is well suited for experimental optimization, in particular for growth media used in cellular agriculture. It also outperforms similar optimization methods, likely due to its ability to synthesize multiple sources of information.
Zachary Cosenza et al. explore the use of a multi-information source Bayesian optimization algorithm to predict an optimal growth media for culturing mouse myoblast cells. Given a commercial growth media with 14 components, they perform optimization experiments using both a traditional design-of-experiment (DOE) method and their Bayesian optimization (BO) method. Cosenza et al. find the BO method reduces the number of required experiments to a greater extent than the DOE method can. They also find that the BO method yields an optimized growth media that more than triples cell proliferation when compared to its commercial counterpart. Overall, Cosenza et al. argue that because the BO method is able to optimize parameters based on multiple sources of information, as is often the case for complicated bioengineering systems like cellular agriculture, it is well-suited for experimental optimization in this field.
Written by Morgan Ziegelski
Cosenza, Z., Astudillo, R., Frazier, P. I., Baar, K., & Block, D. E. (2022). Multi‐information source Bayesian optimization of culture media for cellular agriculture. Biotechnology and Bioengineering.
Andrew J. Stout, David L. Kaplan, and Joshua E. Flack
Dwayne Holmes, David Humbird, Jan Dutkiewicz, Yadira Tejeda-Saldana, Bre Duffy, Isha Datar
Isha Datar, Mirko Betti