Problem:
Therapeutic cells show promise as a new treatment option for chronic illness, but advances must be made in manufacturing reproducibility to get therapeutic cells out of clinical trials and into widespread use.
Project:
Reuel will develop novel, real-time sensors and reinforcement learned (RL), dynamic control policies to improve reproducibility in large-scale manufacturing of differentiated, cell therapies.
The research team will build the reinforcement learning agent and test it using simulations with the latest RL algorithms. In parallel, the team will develop real training environments for differentiated cell culture with multiple sensors and control elements. The training environments will then be used to train the RL agent.
Reuel will benchmark this new, dynamic control approach against static differentiation “recipes” using a range of model cells with Iowa State and external collaborators.
Potential Impact:
Reuel’s work will make possible practical manufacture of regenerative medicines to treat many diseases such as cancers, autoimmune disease, infectious disease, repairing tissue damage and neurological disorders. The new modular framework will be easily tailored to different types of sensors, stimulation cues and cell targets, setting it up to take on emerging biomanufacturing challenges.
“Cell therapies are the next wave of therapeutic innovation, demonstrating incredible outcomes and ability to tailor specifically to each patient. In order to realize their full impact, they need to be made more efficiently and safely. Our work will be a part of this effort,” said Reuel.
Possibilities:
Part of the project includes new course modules, a tool development workshop and even an interactive art exhibit to educate students and the public about machine learning.
“Both graduate students and undergrads will get training and experience in tool design, cell culture and reinforcement learning, further strengthening the nation’s biomanufacturing capabilities,” said Reuel.
2021 NSF CAREER AWARD WINNER
engineering