Technologies such as automation, machine learning, and artificial intelligence (AI) will disrupt the biopharmaceutical industry, helping it overthrow traditional production models in favor of new models that will be consistent with Industry 4.0, the deep integration of the physical and the digital in manufacturing. The Industry 4.0 revolution could bring enormous benefits to bioprocessing, but like any revolution, it must attract committed participants before it has a chance of success.
Christos Varsakelis, PhD, senior manager of processing modeling at GlaxoSmithKline (GSK) in Rixensart, Belgium, says that despite the potential benefits, few in biopharma are as enthusiastic about Industry 4.0 as they should be. “A quick online search reveals that people are quite vocal about their expectations that AI will disrupt drug discovery,” he remarks, adding that these expectations are more modest than they may appear.
“There is a dearth of discussions on the impact of Industry 4.0 on biomanufacturing,” Varsakelis complains. “Yet, many of the foreseeable benefits are straightforward.”
Varsakelis cites real-time monitoring as an area in which AI could have a significant impact. “Accurate predictive-controllers can improve the robustness of the processes and anticipate failure, thereby minimizing waste and guaranteeing product quality,” he points out. “Through continuous monitoring and predictive control, it will be possible to achieve online release, which can substantially affect the supply chain timelines.
“Centralized analytics systems can optimize resource allocation and production schedules. And these systems can consider multiple factors—sustainability, global demand, raw materials availability, weather, etc.”
Other experts maintain that Industry 4.0 concepts have enormous potential for biopharma. For example, Eugene Tung, PhD, executive director of manufacturing IT at Merck & Co., says that analytics platforms will allow biomanufacturers to “improve yield, reduce discards, provide better forecasting of demand, reduce the effort in investigating process deviations, and improve cycle times.”
Digital twins
The idea of creating a “digital twin”—a computer model of a complex manufacturing environment—is another Industry 4.0 idea from which biopharma could benefit. Digital twins were originally conceptualized by NASA in the 1960s to facilitate in silico development, monitoring and predictive control, and real-time asset management.
Digital twins, Varsakelis tells GEN, are exciting for biopharma because their implementation rewards collaboration and benefits from work already accomplished in other industries. “The development of digital twins requires the well-orchestrated collaboration of different departments,” he explains. “For this reason, digital twins are inclusive to a large proportion of employees and, therefore, promote democratization of digital transformation.
“In contrast to most AI initiatives, digital twins constitute a relative mature technology even in highly regulated industries such as the aerospace industry. Biopharma does not need to reinvent the wheel and can accommodate all this experience and know-how.”
Varsakelis adds that digital twins could be combined with augmented reality to improve training: “Operators could be trained virtually even in extreme scenarios at no cost, just as pilots are trained in flight simulators.”
Fact not fiction
Biopharma is notorious for its unwillingness to change manufacturing processes and risk regulatory complications. Earlier this year, a possible explanation for this unwillingness was offered by Janet Woodcock, MD, director of the U.S. FDA’s Center for Drug Evaluation and Research (CDER). “Industry’s reluctance to embrace new technologies,” she suggested, “is probably related to expected regulatory obstacles with FDA and other regulators.”1
“Broad adoption of advanced manufacturing,” she added, “will likely require incentives.”
The CDER, says Woodcock, is working to help biopharma become less risk averse. Specifically, the CDER is identifying emerging technologies, manufacturing processes, control and testing strategies, and platforms that have the potential to advance pharmaceutical quality and modernize manufacturing.
Attempts by regulators to encourage innovation have been noticed. “Regulatory agencies,” Tung observes, “have lauded efforts by manufacturers to improve product quality through the use of data-driven insights and decision making.” Nonetheless, data-driven innovation inspires as much caution as enthusiasm, Varsakelis notes.
“Industry 4.0 challenges us to utilize data in a predictive way,” Varsakelis states. “Although biopharma has been utilizing modeling for years, this has not been done with the objective to allow models to pilot the vessel, but rather to consult the captain.”
“The prevailing view has been that biopharma processes are too complex to be susceptible to accurate modeling,” he continues. “Add on top of that the accumulated risk aversion stemming from the tight regulatory environment, and you get a basic understanding of biopharma’s attitude toward Industry 4.0. It is true, though, that the tide is changing. We hear more and more about concepts such as digital transformation and Industry 4.0. Standard diffusion of innovation theories help explain late adopters. However, whether the reluctance shown by late adopters stems from the overall conservative approach to innovation or from a fear of imminent technological disruption remains murky.”
Industry efforts
Regulatory support is helpful, Varsakelis says, but he maintains that if Industry 4.0 is to be embraced, proponents of the new paradigm will also need to show managers the benefits of digitization. “These benefits are largely qualitative,” he admits, “but one can build upon them and develop quantitative business cases.
“Although innovation finance is not always easy to digest or apply, Industry 4.0 can be nicely approached by a real-option valuation analysis. I personally think that doing so offers a competitive advantage because it allows one to separate fact from fiction and to get a more grounded position on what to expect.”
Implementation
The concept of an Industry 4.0–enabled facility is straightforward. In general, modern equipment is designed to accommodate the installation of multiple sensors. These sensors capture data and enable an Internet of Things (IoT) approach to system management.
However, setting up a facility that uses next-generation technology, such as AI, is very different from establishing a facility that uses older, traditional technologies. “With AI-based solutions in manufacturing,” says Varsakelis. “we have to be comfortable with deploying them before they have reached the desired level of accuracy; in fact, such solutions are deployed so that they may become increasingly accurate.” With AI, newly implemented solutions are almost by definition “not ready,” whereas with conventional technologies, solutions are either ready or not.
Also, a manufacturer can have difficulty determining the range of data it will need for a given process. During facility design, a manufacturer can address this difficulty by following a collaborative approach.
“Whether a company is large or small, a priori assessments of whether intra-corporate data are sufficient to harvest the power of Industry 4.0 are almost impossible,” says Varsakelis. He adds that there are already hints that this difficulty could be overcome if companies and their partners were to participate in data ecosystems.
In a data ecosystem, a company and its partners could share their accumulated data to train superior AI solutions. Each data ecosystem’s whole would be greater than the sum of its parts.
“Understandably, data ecosystems pose confidentiality and IP issues that are prompting uncomfortable discussions,” Varsakelis continues. “As companies such as Google or Amazon get ready to enter the pharmaceutical industry, uncomfortable discussions will intensify.”
Data ecosystems utilize distributed computing systems, which Tung expects to play a greater role in biomanufacturing in the future. He tells GEN, “The best approach to setting up an analytics platform is to leverage cloud-based software that is capable of handling, storing, and processing large quantities of manufacturing data. The biggest challenge is to make the data meaningful to the end user, meaning that the data has to be linked to manufacturing batches in the proper context.”
Retrofitting questions
Implementation challenges may include the need to deal with existing facilities that are kitted out with older systems. “Such equipment can still be subject to upgrades,” says Varsakelis. “However, deciding between upgrading, replacing, or continuing with business as usual should probably be preceded by an appropriate cost-benefit analysis.”
“Many bioprocessing manufacturing environments were not designed with the idea of streaming manufacturing data into an analytics platform as a primary requirement,” Tung notes. This shortcoming, he continues, has been recognized by several industry collaborations, which have “proposed communication and data standards to streamline the sharing of information.” In recent years, various technologies have been developed that are capable of capturing data from established systems and transferring it to centralized analytics tools.
Tung notes that a standardized communication framework can, according to the BioPhorum collaboration, facilitate a plug-and-play approach to the automation of bioprocessing.2 The collaboration has developed a standard data interface between supervisory control systems and single-use biopharmaceutical equipment skids. At present, the collaboration is working to realize true plug-and-play automation for a single-use bioreactor.
Third-party time?
Biopharma companies needn’t develop their own Industry 4.0 capabilities. Instead, these companies can outsource their biomanufacturing to third parties that already possess Industry 4.0 capabilities. However, the outsourcing process can be complex, warns Varsakelis.
“The truth is that the Industry 4.0 market is full of players of different sizes,” he insists. “One has to scrutinize potential partners very well, especially if one is making long-term investments.” He also stresses the importance of forming multidisciplinary teams that incorporate ad hoc experts. These experts, he explains, can “evaluate specifications and requirements and help business intelligence and development reach informed decisions” during negotiations with potential partners.
These negotiations, Varsakelis emphasizes, are not for the faint-hearted. He recalls that he once participated in an Industry 4.0 panel where a technology provider entertained the idea of changing its business model, and accepting the financial implications of doing so, just to further its pursuit of a desirable partnership.
“Along the same lines, we are entering thin-ice discussions on the IP level,” confides Varsakelis. “With all tools being entangled like a Gordian knot, the question of who owns what—or, more important, what can be reused with other customers—becomes highly relevant.”
Facility of the future
Despite the complexity of adopting the Industry 4.0 model, Varsakelis is positive this model will become the norm in the biopharmaceutical sector. He cites the growing use of automation—Industry 4.0 at its most basic—as an encouraging sign.
In manufacturing, interactions between humans and machines occur. These interactions are changing. In industry after industry, machines are acquiring decision-making capabilities. The biopharma industry is no exception. Biopharma, Varsakelis maintains, will adopt Industry 4.0 because the benefits of data-driven manufacturing outweigh any short-term difficulties associated with its implementation.
“It may be too early for us to have a complete view of the factory of the future,” Varsakelis says, “but the concept of a digital factory is anything but farfetched.” He asserts that we can envision production lines that are interconnected with their digital twins, as well as production chains that are automated from end to end. “Owing to the growing tendency for modularity,” he continues, “we can even envision a digital transformation conforming to economies of scale that will augment the potential benefits.”
References
1. National Academies of Sciences, Engineering, and Medicine. Innovations in Pharmaceutical Manufacturing: Proceedings of a Workshop—in Brief. Washington, DC: The National Academies Press; 2020.
2. Tung G, Morris K, Perrone P, et al. The Value of Plug-and-Play Automation in Single-Use Technology. BioProcess Int. 2019.