Industry 4.0, conceptualized by the digitalization-led convergence of digital, physical, and virtual assets, brings a paradigm shift in the totality of operations across the value chain. While many industries are already leveraging the value and benefits of digitalization in manufacturing operations, the biopharmaceutical industries are slowly gearing up in this direction with an industry vision of Biopharma 4.0. Biomanufacturers are investing heavily in building next-generation capabilities toward intensified, connected, and continuous manufacturing to remain competitive and address current and future market needs on speed, quality, agility, value, and sustainability.
Biomanufacturers are taking steps to leverage digitalization, automating the collection, processing, and analysis of data gathered from multiple pieces of production equipment and moving toward Biopharma 4.0 principles. Initiatives are directed at connecting and integrating different operations digitally to provide for centralized data access. Advances in data analytics are affording proactive and predictive capabilities, providing greater process understanding and insights into process deviations and reducing the risk of variability and batch failures while increasing efficiency and quality consistency. Automation systems are reducing manual interventions for greater operational safety and less risk of contamination. There are also opportunities for digitalization solutions related to quality, data integrity, and regulatory compliance.
Irrespective of biomanufacturers’ goals for operational efficiency and building the capability to support next-generation bioprocessing, digitalization is inevitable. Digitalization is a key enabler for such transformational capabilities in biomanufacturing; therefore, the adoption of automation and digital technology is no more a choice but a necessity to remain competitive in the market.
Many benefits of digitalization
Biopharma manufacturing presents enormous opportunities for benefit from thoughtful application of digitalization solutions, according to Matt Panning, senior director of manufacturing at Humacyte. “Controlling product quality, reducing costs, and improving time to market are always important factors in manufacturing. Biopharma manufacturing specifically can often involve small production runs for rare or personalized medicines and/or highly complex processes that require tight controls to ensure product quality and acceptable yield. The costs associated with more traditional infrastructure investment in capital and systems development, as well as the costs of not leveraging the data available from manufacturing, are all very good targets for digitalization solutions,” he observes.
“The benefits of digitalization in biopharma manufacturing are multidimensional across the value-chain of operations. They start with a holistic view of plant operations, facilitating more eyes to control and monitor CPPs (critical process parameters) and CQAs (critical quality attributes) and also include access to centralized contextualized data to gain valuable insights through advanced analytics, such as batch prediction coupled with automated reporting from and across the batch. Real-time access to process data coupled with AI/ML [artificial intelligence/machine learning] [to potentially] support batch release in real-time and significant time savings by minimizing error-prone, manual tasks and enabling easier collaboration across the organization through digital workflows are additional positive impacts. Overall, digitalization helps the organization to achieve operational excellence to meet the market demand on time, speed, quality, agility, and sustainability,” adds Braj Nandan Thakur, global product manager of process automation and analytics for MilliporeSigma, the Life Science business of Merck KGaA, Darmstadt, Germany.
The potential benefits of such digitalization, says Cenk Undey, global head of data and digital for pharma technical development with Roche and Genentech, a member of the Roche Group, include process optimization and continuous improvement to drive cost-of-goods-manufactured (COGM) down and increase staff productivity and safety, while improving environmental sustainability. One of the first improvements, he notes, is the removal of ‘paper’ from the manufacturing process, otherwise required as a system of record. “Digitalization of batch records, equipment-use logs, and such are good examples that help reduce manual data transfers,” he comments.
Operations that heavily utilize process automation are often easy to integrate and mine, but some operations that are often paper-based can be more difficult, Panning remarks. He notes, however, that even some of the paper-based operations can be worth the investment to digitalize if they lead to significant yield or quality improvements through better process understanding and analysis.
Improved process and product insights at a much lower cost result in better manufacturing processes with faster development timelines, contends Jesse McCool, cofounder and CEO of Wheeler Bio, a new biologics contract development and manufacturing organization (CDMO) located in Oklahoma City focused on small-batch, agile drug substance manufacture. For instance, using data acquired in the past combined with dozens of data parameters obtained in real time, it is possible to elaborate through cloud domains and AI which of those parameters can be the best lead or process conditions to obtain a predefined response or target, according to Vincenza Pironti, senior pharma services R&D staff scientist at Thermo Fisher Scientific. “These benefits are applicable for all steps of the process involving all the unit operations present in a biopharma company,” she says.
Digitalization can, overall, impact the very basic philosophy of the biopharma industry and that is access to life-saving therapy in a faster yet compliant manner with proven quality at an affordable price. As a desired outcome of digitalization, while patient safety is of paramount interest, operator safety and sustainability cannot be ignored, Thakur summarizes. “Real-time process monitoring can give insight into process parameters and deviations and reduce the risk of oversight, non-compliance, and batch failures. Predictive analytics can anticipate process deviations earlier and optimize operations to deliver consistent quality. Digitalization can also increase flexibility in biomanufacturing processes by becoming more agile to reconfigure processes and respond to new modalities. Finally, implementing digitalization in biomanufacturing can deliver the goal of faster, improved quality, more cost-effective, safe, and effective therapeutics to patients, irrespective of modalities or steps,” he concludes.
Process development at early clinical stages
During the early clinical development phase, Jesse McCool, cofounder and CEO of Wheeler Bio believes the best place to invest in digitalization is in the process development laboratory where representative (and more affordable) scaled-down models can be leveraged to generate data points in an efficient manner. “Digitalization can aid process scientists in the ML cycle where we design, test, build, learn, and repeat, in an iterative fashion, to create powerful, predictive models of our biomanufacturing processes. The data wrangling is perhaps the hardest part when implementing ML in the lab and digitalization solutions are key [to] reducing the manual steps in data processing,” he explains. The process is iterative in nature, so according to McCool, digitalization tools are most useful at this stage to keep costs down and expedite insight building.
Drug development can also benefit from various aspects of digitalization, such as lead definition through the use of AI, according to Vincenza Pironti, senior pharma services R&D staff scientist at Thermo Fisher Scientific. She notes that computational chemistry can be used as well to broaden the pool of potential candidates and explore non-standard landscapes, thus expanding pipelines. “In summary, three values are widely acknowledged as the major benefits of digitalization at early development phases: high productivity, possible timing reduction in drug discovery, and diversified pipelines,” Pironti concludes.
Upstream and downstream nuances
There are other factors that can impact the level of benefits realized through digitalization. There may be process-specific nuances that differ between upstream and downstream operations, Thakur observes. “For example,” he explains, “if one looks upstream, where the harvest parameters are crucial to the downstream outcomes, there is the added challenge of accommodating the unknown variations during the process. While some variations can be measured, for others the real impact may not be known until later in the process. This introduces the need for constant measurement through the duration of a run which can span multiple days, during which significant online data [are] generated along with the offline data for the period of the run.”
Digitalization can help here, Thakur comments, because it enables monitoring of variability due to different sources—the genealogy of the raw materials used and continuous reviews of the quality profile (such as the glycosylation profile and charge distribution for antibodies) using advanced sensor technologies, process data analytics capabilities, and advanced analytical models.
Another big difference between upstream and downstream is referred to as the process time constant, such as in biologics drug substance manufacturing, according to Undey. Downstream operations are much faster and end more quickly—sometimes a phase lasts a minute versus weeks—for a process in a bioreactor. “In addition, while the types of data collected are very similar, more types tend to be evaluated for quality control (QC) samples obtained downstream for quality attribute analysis for lot release compared to those for upstream in-process control (IPC) samples collected for process performance variables,” he says.
Upstream, there can be further differences in digitalization for batch/fed-batch processes versus intensified processes performed in a smaller footprint. Integrated continuous manufacturing (CM), meanwhile, requires fully connected data flows between upstream and downstream operations (such as the harvest, chromatography, and ultrafiltration/diafiltration steps) with surge tanks as well as process analytical technology (PAT) sensors for ensuring process and product quality control across the CM process operation, according to Undey.
New approaches to data analytics software
Many vendors and tools exist today that make data analysis of biopharma manufacturing data and supply-chain metrics easier to understand more quickly. One trend is the availability of scaled licenses for software-as-a-service that allow minimal investment up front and flexibility to adapt while also giving manufacturers a robust package of tools that meet the expectations of their patients, regulators, and investors for delivery, cost, and quality, according to Matt Panning, senior director of manufacturing at Humacyte.
“While the heart of novel drug discovery and development may always rely on human ingenuity at its core, ensuring that therapeutics are produced consistently and safely through the use of digitalization services can be very advantageous,” Panning contends. Process analytics software, web-based ERP systems, and even off-the-shelf production-scheduling software that makes it easier to coordinate the myriad activities involved in current good manufacturing practice (CGMP)-compliant manufacturing are just a few examples he points to.
Overcoming challenges
As with any major change in operational activities, there are hurdles that companies must overcome when moving to digitalized bioprocessing. One of the first that must be overcome is linked to natural human resistance to change. Steps that can be taken to minimize this issue, says Pironti, include creating a new mindset inside the organization to build necessary skills in the workforce and new roles for monitoring the progress and the outcomes resulting from the introduction of digital tools. Performance of gap and risk assessments should help identify specific needs and enable development of appropriate action plans.
Another human challenge noted by McCool is found in the intellectual interface between process scientists (subject matter experts) and data scientists (modeling experts). “I think it is quite difficult for most data scientists to contextualize bioprocess data without having a deep knowledge and tangible experience of the real-world physical systems they are modeling. Therefore, it is crucial to have a very good collaborative approach for model-based operations where process scientists and data scientists work together on a regular basis. Additionally, this structure needs executive sponsorship at the highest levels in the company for a disciplined ML strategy to emerge and be sustained,” he explains.
On the technical side, companies can find it challenging to select the right digital technologies and integrate them into their ecosystems, according to Thakur. “An existing facility comes with the challenge of integrating digital technologies into its legacy systems and existing infrastructure, while a new facility can build a new solution but faces the dilemma of choosing already-known technologies instead of innovative ones to reduce the risk and complexities of validation,” he says.
Adding to this difficulty is the lack of a unified out-of-the-box digital solution that applies across the biomanufacturing workflow and universal standards for software interfaces, which means biopharma companies must work with multiple vendors and systems to ensure that all these digital technologies are compatible and can be integrated together. “Standardization across the organization and across sites globally can be challenging as local digitalization needs can differ from global needs. When digital technologies are implemented in pockets of an organization or as a pilot to demonstrate value, the challenge remains in scaling up digital technologies to the wider enterprise. Last-but-not-least is the challenge of articulating the value realized from investments in digital technologies,” Thakur observes.
Lack of ubiquitous data standards across the industry, adds Undey, was confirmed by the domain experts from industry, academia, and suppliers during the Integrated Continuous Biomanufacturing conference that took place in Sitges, Spain in October 2022. “Different manufacturing technologies often have different data standards, and therefore, data integration requires further investment on the part of the biopharma company. Additionally, manufacturing digital systems are typically not plug-n-play, so each time we introduce a new piece of equipment or analyzer/sensor, effort is required to make its data integrated with the rest of the labs and plants,” he says.
Lack of standardization raises other questions as well. Assuring data integrity can be challenging while digitalizing large amounts of data from different sources and integrating them across multiple platforms. “Despite being an advantage overall, the use of multiple service providers who each deliver and manage their own solutions to a manufacturer can present the manufacturer with a significant problem when it comes to ensuring that all of the data maintains integrity across all of those platforms,” Panning comments.
In addition to controlling primary data sources, there are issues around handling the impact of software updates on data validation. “Manufacturers, not their service providers, are liable to understand and control their data collection, processing, and retention are maintained when these routine updates happen. And when two or more of these services are linked, which is often the case when a manufacturer sets up their own customized solutions, it makes keeping up with all of the changes that much more difficult,” Panning explains. The rate of change across the industry and regulatory agencies further heightens such concerns.
Accessing comprehensive data is another important issue facing biopharma companies looking to digitalize their manufacturing operations. According to McCool, Michelle K. Lee, founder and CEO of Obsidian Strategies and former AWS and Google executive, remarked in a MLxBio5 conference in San Francisco on Oct. 21, 2022 that companies are really only using approximately 10% of the data that gets generated (1). “The real work for any company looking for digital transformation is getting access to all the data being generated in their company,” McCool contends.In addition, he notes that more than half of the time spent developing ML models is spent wrangling, processing, and cleaning up data—activities that are ripe for automation by smart middleware solutions.
Furthermore, the biopharma industry, McCool observes, is not yet data-rich like other industries, even for commercial production, let alone clinical operations. “Bioprocesses are fraught with complexity, real-world processing interactions, and regulatory constraints that make predictive modeling fundamentally challenging. When this truth is considered along with the fact that data points are very expensive in bioprocessing ($6–$10 million for a Phase I manufacturing batch), it is obvious why data strategies currently remain limited,” he states. He reiterates that the solution, at least in part, is to have an accurate scaled-down model of bioprocesses coupled with automated workflows and smart middleware that lowers the cost of collecting and analyzing data points.
Different approaches but some common best practices
Different companies take various pathways toward digitalization. Some focus efforts on new facilities, building “digital plants” or at least digitalizing to the fullest extent possible certain production suites. Others seek to implement digital tools and technologies within existing operations for legacy products to gain experience and determine the level of benefits that can be realized in practice before expanding to wider adoption.
“Digital transformation is a journey that can occur in various parts of an organization, across a department, a site, or multiple sites globally, or with suppliers and collaborators,” Braj Nandan Thakur, global product manager of process automation and analytics for MilliporeSigma, the Life Science business of Merck KGaA, Darmstadt, Germany observes. “There is not a one-size-fits-all implementation strategy, but there are some common approaches to ensure that the journey to digital transformation is successful,” he says. First, it is important to start by evaluating the organization’s current digital maturity and understanding the vision and future state. Then, it is essential to evaluate the value of digitalization to the organization and build a well-thought-out and tailored overarching strategy and roadmap that fits the organization’s needs. “The technology landscape is complex and may be confusing; however, laying a robust foundation on the digital journey considering current business challenges and anticipated future challenges is key to long-term success. Hence, it’s recommended to adopt a solution-centric phased approach that minimizes the risk and maximizes the value,” contends Thakur.
Deciding what areas to digitalize and for what purpose and whether to digitalize end-to-end or if it is necessary for the main objective is a first step, according to Cenk Undey, global head of data and digital for pharma technical development with Roche and Genentech, a member of the Roche Group. Once this is understood, data standardization and technology platform selection (whether to invest in data lakes and meshes, IIoT platforms, advanced analytics, and visualization) are determined. “Probably the most crucial aspects of the strategy are establishing necessary processes and training the workforce to sustain and enjoy digital plant operation. This may involve upskilling the existing workforce at all levels for digital literacy and hiring new talent with advanced skills (like data/computational/ML scientists as necessary),” he notes.
Panning agrees that establishing some fundamentals is important. That includes developing a clear set of policies on data integrity, change management, and supplier management (including supply and quality agreements). “These tools will make it easier for their organization to identify and develop digitalization solutions in a more consistent manner,” he explains. Communication is equally important, according to Panning, and all relevant functional groups in a company should be kept in the loop throughout implementation.
It is also critical, stresses Jesse McCool, cofounder and CEO of Wheeler Bio, to use data that are important for the company and to have a comprehensive data management strategy that is supported by the highest level of corporate governance. “The ML model is only as good as the data it is trained on, so having access to meaningful data is absolutely essential,” he states. That requires developing a good strategy within which data are viewed as an organizational asset and not a departmental asset, thus ensuring that data are democratized easily and securely so that they can be leveraged for analytics and ML applications.
“Implementing digitalization strategies requires a mindset and cultural shift and change across business and operational models and organizational structures. A culture of innovation, collaboration, and continuous learning through proof of concepts and pilots allows the organization to evolve its digital maturity and learn how to maximize its value of digitalization upon scale-up,” Thakur concludes.
Current state of digitalization
The level of adoption of advanced digitalization in the biopharma sector is fairly low, but that is changing. Most mid-to large size biopharma companies, says Undey, digitalized batch records in the early to mid-2000s to remove, or significantly reduce, paper records from manufacturing floor operations, although old paper-based recordkeeping may still exist in various spots due to organic growth and investment prioritization reasons. Process automation that integrates widely across a site and use of data analytics to better understand and control processes is expected at this point, according to Panning. He also notes that the use of service providers for licensed software has exploded in the past decade or so.
Even so, Panning believes the industry is still grappling with how to apply more traditional pharmaceutical models to the new models that are evolving today. In addition, McCool believes that currently manufacturers are making do with 21st-century digital tools and automation that are not sufficient. “To really penetrate into the heart of what is going on with their systems and achieve deeper understanding than is currently possible, process scientists need predictive modeling, digital twins, and AI augmenting their abilities,” he insists.
Full data integration across the plant is rare, Undey agrees. “A cross-industry forum led by BioPhorum group has created a maturity model for biomanufacturing digital plants to gauge the need and the level of digitalization and automation. The study found that for most plants, ‘connected plant with data integration’ is sufficient, while some fully autonomous/adaptive plants may make more sense to drive COGM down, depending on the production volume and staffing considerations,” he notes (2).
The COVID-19 pandemic has, however, put pressure on biopharma companies to pursue digitalization more quickly. “In an unprecedented way, biopharma companies, industry stakeholders, and regulatory agencies collaborated to accelerate the adoption of digital technologies for faster development and manufacture of vaccines and therapeutics and to facilitate day-to-day operations,” Thakur says.
As a result, today large biopharma companies look at digitalization as a transformation lever that helps them keep the core knowledge, architecture, and designs internally while working with strategic partners to execute the tactical technology implementation, according to Thakur. Smaller firms, however, often adopt digital capabilities that help them solve operational and compliance needs in the short-medium term while giving them a competitive edge in terms of operational efficiency, business intelligence, and customer collaborations so they can meet aggressive timelines. Contract service providers leverage digitalization strategies that afford them greater flexibility and scalability and the ability collaborate with their customers and regulators.
The pandemic, adds Pironti, prompted an incredible acceleration in digitalization of biomanufacturing—from lead selection to the generation of prediction modeling for the manufacturing process, to prevention of possible deviations through monitoring storage and distribution data. “These digital assets can be the starting point for further implementation of digital tools,” she observes.
Indeed, Undey underscores the fact that several companies have been making steady progress with digitalization of their bioprocessing operations. Example solutions he highlights include plantwide real-time process monitoring, predictive maintenance of equipment, PAT, cobots for warehouse ops, various “Industry 4.0” applications such as the Industrial Internet-of-things (IIoT), integrated MES/ERP/laboratory information systems (LIMS), applications of AI and ML, and advanced process control solutions. “The adoption of these technologies is driven by their technological maturity and availability in the marketplace (with robust suppliers), regulatory acceptance and experience, and ability to meet key business needs,” he states.
Biopharma moving from adopter to pioneer
Digitalization is an exciting journey that continues to evolve within the biopharma industry as manufacturers continue to improve performance and strive for greater efficiency and quality, all with the goal of meeting current and future patient needs. “We should remember that digitalization of biomanufacturing is a means to that end to help drive utmost efficiencies and COGM down, ensuring supply to patients,” Undey emphasizes.
Areas within digitalization where further developments can be expected, according to Undey, include further capabilities for remote monitoring and support technologies and more advanced applications such as miniaturized wireless sensors, machine-learning programs for predictive plant-performance management, intelligent decision-support systems, and digital twins. “It is not a matter of if, but a matter of when. We are excited to help create that future and embrace it to harness the power of data and digital across the product-advancement value chain to better serve our patients,” he states.
Within change lie both opportunity and challenge, and that is true for digitalization of biomanufacturing, Panning observes. “Digitalization has enormous potential for our industry and will only continue to grow in its applications. Picking the right solutions, at the right time, and following through with proper resourcing during the change-management process are the keys to successful implementation,” he comments.
The adoption of digitalization in bioprocessing is, Thakur believes, happening at a very rapid pace, with Industry 4.0 technologies evolving at an even more rapid pace. The challenge for biopharma manufacturers is to keep themselves in step with this transformation. “For now, biopharma is more of an adopter than a pioneer in these applications. That, however, is going to change in the coming years. With the advancements in precision medicine, new modalities, and novel therapies like cell-and-gene to name a few, the space of biologics is going through a change that will demand a lot more precision and accuracy from these digital technologies that other industries may not demand of them,” he says.
The reason is simple, according to Thakur—biopharma touches lives like no other industry, and as the space of biologics starts rapidly evolving, the demand for innovation and technology will also be unprecedented. “This pace cannot be met by biopharma manufacturing organizations alone. Instead, it will require an ecosystem of technology companies, digital providers, manufacturers, regulators, governments, entrepreneurs, and life-science leaders alike, who will come together to transform the way biopharma will evolve in the coming decade. A lot more is yet to come, and this is just the beginning!”
References
- M. Lee, Obsidian Strategies, “Seven Lessons for Successful Machine Learning Deployments,” Presented at the 5th Annual Machine Learning Meets Biology MLxBio5 Conference, San Francisco, Oct. 21, 2022.
- E. Anttonen et al., “The Development of a Digital Plant Maturity Model to Aid Transformation in Biopharmaceutical Manufacturing,” White Paper, BioPhorum Operations Group.
Article details
BioPharm International
Vol. 35, No. 12
December 2022
Pages: 24–33
Citation
When referring to this article, please cite it as C. A. Challener, “Digitalization of Biomanufacturing: A Status Update,” BioPharm International 35 (12) (2022).
