The biopharmaceutical industry is poised for a digital transformation. For Tiago Matos, Associate Director of Bioprocess Drug Substance and Commercialization at Merck & Co., this shift is not just a matter of adopting new technologies.
It is about rethinking processes, strategies, and mindsets to align with Industry 4.0's potential. Although the sector remains rooted in systems developed during earlier industrial revolutions, Matos believes the time is ripe to connect silos and harness technologies such as continuous manufacturing, digital twins, and machine learning.
Matos argues that the field has yet to truly transition from Industry 3.0, which was built on computer-based automation, into the fully integrated cyber-physical systems envisioned by Industry 4.0. Biologics manufacturing lags, but the foundations for transformation are forming. According to Matos, being the first to adopt every new technology will not be about being the first. Instead, the goal is to pave a path that others can follow, making the ecosystem stronger as a whole.
What drives Matos is not only a professional mission but a deeply personal one. His early experiences with asthma in his family sparked a lifelong curiosity about genetics, immunology, and chronic diseases. That curiosity led him to biochemistry, then process engineering, and finally biomanufacturing. Each step of the way added a new dimension to his understanding of how science and technology can intersect to improve lives.
This concept is discussed in greater detail in an episode of the Smart Biotech Scientist Podcast, hosted by David Brühlmann, founder of Brühlmann Consulting.
Industry 4.0 in Biopharmaceuticals: Reality Check
Understanding the Gap
Matos offers a candid assessment: biologics manufacturing is behind other sectors in adopting Industry 4.0. Despite the official landmark of Industry 4.0 in 2010, most manufacturing operations still reflect the logic and systems of the 1970s. While digital systems have enhanced some batch operations, truly integrated processes remain rare. The transition will require new tools, organizational collaboration, standardization, and shared goals.
Not Just a Buzzword
Industry 4.0 is more than a hype cycle. It represents a tangible direction for the industry, involving technologies such as AI, the Internet of Things (IoT), and cyber-physical systems. These tools can seamlessly link development, scale-up, and commercial manufacturing. However, the road ahead requires harmonization across companies and regulators to define shared standards and strategies.
Emerging Roles for Collaboration
Companies like BioPhorum are helping drive that collaboration, providing platforms for conversation and consensus. The more industry leaders agree on shared roadmaps, the faster the transformation unfolds. However, the end destination may look different from today's vision. As Matos puts it, the industry is learning in real-time, and the shape of Industry 4.0 a decade from now will reflect that evolution.
Continuous Manufacturing and Digital Strategy
Continuous Control Strategy
While continuous manufacturing is often touted as the future, Matos emphasizes a nuanced perspective. The real leap is in the control strategies that support process operations. Even if fed-batch remains the dominant model, improvements in silico tools, more expansive design spaces, and holistic control strategies can bring the benefits of flexibility and faster decision-making.
Holistic Process Thinking
Instead of focusing on one unit operation at a time, integrated modeling tools enable end-to-end oversight. If a bioreactor drifts from set parameters, teams can respond downstream without discarding the batch. These capabilities are enabled by combining mechanistic process knowledge with predictive models and machine learning, even when real-time analytical data are limited.
I think that holistic control strategy is going to be a fundamental key change here and connecting all of these new tools. Basically more realistic mechanistic understanding of processes better in silicon models. But as well where we don't have actually analytical tools that provide you that real time test release machine learning algorithms can actually give you that fundamental knowledge, can provide you the hint of what's happening.
The Role and Limitations of Digital Twins
Where Digital Twins Fit
Digital twins can potentially transform process development and scale-up, but Matos cautions that they are not one-size-fits-all solutions. Their use makes the most sense in platform processes, where workflows are repeated and predictable. For bespoke products or highly variable molecules, the cost and complexity of developing a digital twin may not be justified.
Clarifying the Concept
There is also a widespread misunderstanding about what a digital twin truly is. A mechanistic model alone does not qualify. A complete digital twin integrates real-time process automation, analytics, and simulation models to provide dynamic control. This combination allows immediate operational adjustments and predictive control strategies, forming the next generation of advanced process control.
Cost and Expertise
Digital twins are expensive and resource-intensive. They require automation infrastructure, analytical sensors, robust models, and the ability to integrate these systems. Multiple subject matter experts must collaborate to build and validate these systems. The investment is significant, but so is the potential payoff for the right applications.
Small Wins, Big Impact
Proof-of-Concept Projects
To build momentum, Matos and his team focus on small proof-of-concept applications. These include early prediction of membrane fouling and resin performance decay in chromatography. Using predictive models, they can adjust parameters such as transmembrane pressure to prevent issues before compromising product quality.
Value in Continuous Chromatography
As continuous chromatography gains popularity, a deeper understanding of resin dynamics offers clear advantages. Predictive modeling enables more accurate loading and longer resin lifespans. These wins build confidence and establish value before the broader adoption of full digital twins.
Vision for End-to-End Models
Ultimately, Matos wants to design facility-fit models that replace static mass balances with dynamic simulations. If process deviations occur, these models will enable teams to trace issues back upstream in real-time. This ambitious goal must proceed incrementally through validated, cost-effective milestones.
Regulatory Landscape and Validation
A Proactive Stance from Regulators
Contrary to common perception, regulatory agencies are not a barrier. Matos sees them as collaborative partners who welcome early engagement. Agencies are eager to understand new approaches, especially when supported by sound data and well-prepared briefings.
Key Validation Challenges
The main challenge is determining where digital twins fall in the regulatory framework. Are they part of automation? Are they models requiring separate validation? Because they span both domains, teams must validate the underlying models and the automation systems they connect to.
Business Considerations
Beyond technical hurdles, companies must also evaluate cost and scalability. Off-the-shelf software options exist but come with licensing fees and ongoing maintenance. Custom internal solutions offer flexibility but require sustained investment in skills and infrastructure. These considerations impact the feasibility of scaling digital strategies across multiple sites.
Machine Learning and the Future of AI
Validation of AI Tools
Machine learning and AI models are growing in popularity, but they introduce new validation challenges. While models can be trained and tested on lab data, regulatory guidance is still evolving. Currently, the field lacks a formal pathway for validating cognitive or adaptive systems in live manufacturing environments.
Role of Soft Sensors
One promising area is using soft sensors to bridge gaps in analytical capability. For example, Raman spectroscopy combined with data modeling can estimate parameters not directly measured in real-time. This can offer a richer, more responsive control environment without requiring perfect instrumentation.
Not All PAT is Enough
Despite advances in Process Analytical Technology (PAT), some critical quality attributes remain difficult to monitor in real-time. Digital systems and machine learning can complement PAT by filling in gaps and providing surrogate quality indicators. This layered approach is crucial for smart manufacturing environments.
I think continuous will be the future to come near soon. And then automation will start becoming even more easier to implement. Advanced controls, model based controls, some AI driven machine learnings that will help us taking a little bit of lesson learned during campaigns without having that true control over the scheme, of course.
Envisioning Smart Manufacturing
A Long Road Ahead
According to Matos, fully smart manufacturing is still 15 to 20 years away. It requires integrating closed-loop systems, real-time release testing, and digital design frameworks that can predict performance based on molecular inputs. These elements are still in the early stages of development and adoption.
Key Milestones
Some critical milestones include:
- Harmonization of platforms and control systems
- Development of plug-and-play unit operations
- Creation of end-to-end process design tools
- Widespread use of wireless, modular equipment
These changes depend on technological maturity and capital investment. Given the high cost of equipment and batch production, companies are cautious about adopting new systems until they clearly demonstrate value.
Collaboration and Skills for the Future
No One Can Do It Alone
A core message from Matos is the importance of collaboration. At Merck, he works closely with colleagues from modeling, automation, and process development to form what they call the Digital Trio. This diverse set of skills and perspectives is essential for innovation.
Blending Disciplines
Scientists entering the field are encouraged to develop fluency across multiple domains. Even if someone is not a modeler or automation engineer, they should be able to communicate effectively with those teams. An interdisciplinary understanding can make the difference between an idea and a successful implementation.
Lifelong Learning
Matos credits his mentors and international experiences with shaping his approach. He embraced opportunities to learn new tools and perspectives, from Portugal to Sweden to the United States, and that openness continues to fuel his passion for pushing boundaries in biomanufacturing.
Final Remarks
Tiago Matos remains optimistic yet realistic about the path forward. Industry 4.0 and digital twins are not easy solutions, but they are critical enablers for the next generation of biologics manufacturing. The industry can bridge the gap between ambition and reality through collaboration, strategic investment, and regulatory alignment.
The road is long, but the potential is extraordinary. The digital transformation of bioprocessing is already underway for those willing to embrace complexity and build bridges across expertise areas.
About Tiago Matos
Associate Principal Scientist/Associate Director in Bioprocess Drug Substance and Commercialization at Merck & Co. with over 10 years in the pharmaceutical industry. With a strong record of publications, patents, and international awards, focusing on biologics and vaccines, with an emphasis on process development and smart manufacturing. Leveraging a strong biochemistry background, target to enhance process design and to develop control strategies in the digital era.
Currently leads a team dedicated to implementing next-generation technologies, aiming to transform drug development and commercialization by advancing process efficiency and robustness through digital innovation, like models and digital twins.
Connect with Tiago Matos on LinkedIn.
David Brühlmann is a strategic advisor who helps C-level biotech leaders reduce development and manufacturing costs to make life-saving therapies accessible to more patients worldwide.
He is also a biotech technology innovation coach, technology transfer leader, and host of the Smart Biotech Scientist podcast—the go-to podcast for biotech scientists who want to master biopharma CMC development and biomanufacturing.
Hear It From The Horse’s Mouth
Want to listen to the full interview? Go to Smart Biotech Scientist Podcast.
Want to hear more? Do visit the podcast page and check out other episodes.
Do you wish to simplify your biologics drug development project? Contact Us
