Drug companies looking to intensify output need to embrace artificial intelligence, says the author of a study suggesting tech can overcome the challenges involved in continuous-mode production. For a biopharmaceutical company, the potential benefits of continuous manufacturing are well established and worth the investment, according to Dongming Xie PhD, from the department of chemical engineering at the University of Massachusetts.
“A successful continuous manufacturing approach may result in several-fold increases in the volumetric productivity while keeping similar product titer and conversion yield,” he notes. “This will significantly reduce the operating cost and capital investment, adding that there are also potential product quality benefits.”
“In addition to the advantage in reducing the operating cost and capital investment, continuous biomanufacturing may also lead to more consistent and high-quality final product due to the use of steady-state process conditions in the continuous manufacturing process.”
But drug firms used to traditional batch-mode production must keep in mind that 24/7 manufacturing presents unique technical challenges continues Xie, who looked at the issues in article published this month.
“There is a higher contamination risk with microbes from the environment during the long-period operation of a continuous process. Also, there can be issues related to the genetic instability of strain, [i.e.] the cell line, which causes significant loss of cells’ specific productivity,” he points out.
Addressing these challenges formerly involved a significant amount of experimentation and trial and error. In the past few years, however, the biopharmaceutical industry has turned to computer science tools.
“Artificial intelligence (AI) and machine learning have recently been applied to biomanufacturing. The integration of AI into continuous fermentation process allows for the identification of critical factors impacting the fermentation performance, application of real-time feedback control, and maintaining process reliability with timely adjustments and controls for variations or fluctuations from strain growth characteristics, medium/feed ingredients, and process operations,” Xie says.
Xie cites interpretable modeling, representation learning and graph attention networks, reinforcement learning, and deep reinforcement learning as examples of AI techniques industry is relying upon to monitor and understand culture conditions.
“Such tools can be used to increase efficiency, robustness, reliability, and yield with significantly lower manufacturing cost,” he tells GEN. “Continued research is still needed to determine the most appropriate AI tools for continuous biomanufacturing.”
Another, potentially even more critical requirement for a successful continuous manufacturing setup, is having bioprocessing technologies that monitor and report process data to a centralized control hub in real time. according to Xie, who explains that the “availability of high-quality process data from industry is critical so that AI can be more successfully applied to the biomanufacturing areas.”