NSF CAREER Award for Generative AI-Driven Bioprocess Learning and Optimization To Advance Biomanufacturing
MIE Assistant Professor Wei Xie was awarded a $596,920 NSF CAREER Award for “Mechanism-Informed AI for Biological Systems-of-Systems to Accelerate Biomanufacturing Systems Integration and Innovations.” The project aims to create innovative bioprocess-specific AI for general biological ecosystems that can facilitate flexible and intelligent on-demand manufacturing systems for biopharmaceuticals.
Wei Xie, assistant professor of mechanical and industrial engineering, received a $596,920 National Science Foundation CAREER Award to develop a new mechanism-informed AI platform that can lead to transformative innovations in highly complex biopharmaceutical manufacturing by accelerating manufacturing systems integration, development, and automation with dramatically improved capabilities and flexibility.
The goal is to create a unified knowledge representation characterizing spatial-temporal causal interdependencies of biomanufacturing processes from molecular to cellular and macroscopic scales. This knowledge representation enables the integration of sparse and heterogeneous data collected for different manufacturing systems from lab-based to large-scale industrial manufacturing.
Further, in conjunction with innovative federated and optimal learning, the generative AI platform introduced by Xie’s team for biological systems-of-systems enables a quick assembly of flexible, robust, and intelligent biomanufacturing systems with dramatically improved capabilities and responsiveness in the production of existing and new drugs. This is particularly relevant to producing personalized biotherapeutics treatments and iterations of vaccines to address rapidly mutating viruses.
Unlike traditional small molecule drugs, biopharmaceuticals, also known as biological medical products, use living organisms (such as bacteria, yeast, and mammalian cells) as factories that provide essential life-saving treatments for severe and chronic diseases. These include cancers, autoimmune disorders, metabolic diseases, genetic disorders, and infectious diseases such as COVID-19. Unlike traditional medications and drugs, biopharmaceuticals often have increased efficacy and reduced side effects.
But the use of living organisms can be unpredictable, particularly when interacting with other species, organisms, and new environments on a cellular or molecular level. Biomanufacturing processes can have hundreds of biological, physical, and chemical factors dynamically interacting at multiple levels. With very limited, sparse, and heterogeneous data, it is often hard to quantify exactly what occurs during these production processes.
To respond to this complexity and uncertainty, biopharmaceuticals manufacturing needs to be highly controlled and regulated and as a result, it is expensive, and it takes a long time to get products to market. “Existing separated manufacturing systems and good manufacturing practice, built on pre-specified process parameters, have a lot of limitations for handling the challenges of biopharmaceutical manufacturing, such as its high complexity and the high uncertainty with very limited data” Xie says.
Xie’s aim is to enhance biomanufacturing systems integration and intelligence through creating innovative generative AI for biological systems-of-systems that will advance scientific understanding and guide sample efficient learning from limited heterogeneous data collected from different production processes at multiple scales. She plans to leverage emerging sensing technologies to monitor bioprocessing at molecular and cellular scales, so this new AI platform can quickly decode and capture what is happening during the production process.
Continuous and active learning will enable more flexible assembly of a new biomanufacturing system and drug production.
“We can leverage all historical data and information and accelerate and scale up future intelligent biomanufacturing systems,” Xie says. “By doing so, many people will benefit.”
In the case of vaccine production for mutated viruses, “we will leverage the information on the major parts of the new virus that have not changed and then guide sample efficient and active learning to advance the scientific understanding on the part of the virus that has mutations,” Xie says. “Using that information can guide the optimal design of experiments focusing on the mutations and streamline the production processes of the new vaccines.”
The research could impact life-saving treatments for diseases such as diabetes and certain cancers, while also improving treatments for chronic illnesses like rheumatoid arthritis.
Longer term, these new modeling and AI methodologies could be applied to other manufacturing processes, including the manufacturing of biofuels and food products, and expand to healthcare and environment sustainability,” Xie says.
Abstract Source: NSF
See related article: Northeastern researcher receives CAREER award to advance biopharmaceutical manufacturing with AI