Konstantinos Alexias

Scientist, API Process Development, Biotherapeutics

Konstantinos Alexias has a background in bioprocess engineering with expertise in upstream and downstream biopharmaceutical process development, technology transfer, and digitalization. Holding a Professional Doctorate in Engineering (EngD) from Delft University of Technology and a Chemical Engineering Diploma from National Technical University of Athens, Greece, Konstantinos specializes in the early and late-stage development of biopharmaceutical manufacturing processes, computational modeling, process intensification, and PAT integration. As a Scientist in API Process development of Advanced Therapies Leiden at Johnson & Johnson his work includes process optimization, CFD modelling and machine learning applications in bioprocessing. At DataHow Symposium 2025 @ ETH Zurich, he will share insights on AI-driven bioprocessing, digital transformation, and the future of pharmaceutical manufacturing.

Hybrid-modeling in mRNA manufacturing process development: An industrial case study

One of the most important and widely used process steps in mRNA drug substance manufacturing is the In Vitro transcription reaction. This process step allows for template-directed synthesis of RNA molecules of any sequence from short oligonucleotides to those of several kilobases long.

Despite its mechanistically straightforward nature, the optimization of the In Vitro transcription reaction for mRNA manufacturing processes can be quite challenging. In view of the ever-growing capabilities of hybrid models in biopharmaceutical process development, a study was conducted to address this optimization challenge by developing a hybrid-model to characterize and sufficiently predict the performance of this process step.

Prof. Antonio del Rio Chanona

Associate Professor

Antonio del Rio Chanona is head of the Optimisation and Machine Learning for Process Systems Engineering group at the Department of Chemical Engineering, and Sargent Centre, Imperial College London. He is Director of Education at the Sargent Centre for Process Systems Engineering, and Co-director of the Centre for Doctoral Training in Next Generation Synthesis & Reaction Technology (rEaCt).

Antonio’s main research interests include Data-Driven Optimisation, Reinforcement Learning, Control and Hybrid Modelling applied to process systems engineering.

Antonio received his MEng from UNAM in Mexico, and his PhD from the University of Cambridge where he was awarded the Danckwerts-Pergamon Prize for the best doctoral thesis of his year. He received the EPSRC fellowship to adopt automation and intelligent technologies into bioprocess scale-up and industrialization and has received awards from the International Federation of Automatic Control (IFAC), the Institution of Chemical Engineers (IChemE) and the the Association of European Operational Research Societies (EURO) in recognition for research in areas of process systems engineering, industrialisation of bioprocesses, and adoption of intelligent and autonomous learning algorithms to chemical engineering.

Antonio is very enthusiastic about teaching and mentoring, he is the Director of Education for the Sargent Centre, has organized several summer schools included The Data Driven Optimization Summer School (2022), The mathematical and statistical foundation of future data-driven engineering (2023), Bayesian Optimisation Summer School (2024), amongst many others.

From human-in-the-loop to LLM-in-the-loop Bayesian optimisation

Bayesian optimization has proven effective for optimizing expensive-to-evaluate functions in design of experiments (DoE). However, valuable insights from domain experts are often overlooked. This work introduces a collaborative Bayesian optimization approach that re-integrates human input into the data-driven decision-making process. By combining high-throughput Bayesian optimization with discrete decision theory, experts can influence the selection of experiments via a discrete choice. We propose a multi-objective approach to generate a set of high-utility and distinct experimental designs, from which the expert selects the most promising one for evaluation at each iteration. Our methodology maintains the advantages of Bayesian optimization while incorporating expert knowledge and improving accountability in experimental planning.

Beyond human expertise, we further explore replacing domain experts with multi-agent large language models (LLMs) that simulate expert decision-making. These agents, assess candidate experimental designs, enabling an automated yet interpretable alternative to human-guided DoE. By leveraging a team of specialized LLM agents, we investigate their effectiveness in balancing trade-offs, capturing diverse expert opinions, and accelerating decision-making. Through case studies, including bioprocess optimization and experimental design for new material discovery, we demonstrate that both human and AI-driven expert guidance enhance experimental planning. By continuously incorporating either human or AI-driven expert opinion, the proposed method enables faster convergence and improved accountability for Bayesian optimization in design of experiments.

Dr. Tanja Hernandez

Senior Expert – DSD Modeling and Simulation Lead

Towards model-informed drug substance process development – The role of prior knowledge and decision-making under uncertainties

Dr. Dietmar Lang

Senior Director Platform Development Program Management

Living the digital transformation in a small biotech company

The introduction of digital twins in the pharmaceutical industry marks a significant leap towards Pharma 4.0, particularly for biotech companies like CureVac.

Digital twins, which are virtual replicas of physical entities, enable real-time simulation and analysis of mRNA, antibody and other drug development processes. This technology, in combination with real time analytics, will enhance quality control for the process and product, will facilitate the optimization of production processes and will consequently accelerate time-to-market for new drugs / drug modalities. The implementation of digital twins aligns with the principles of Pharma 4.0, which emphasizes the integration of digital technologies into all aspects of pharmaceutical manufacturing. Leveraging these tools might enhance, streamline and accelerating bioprocessing, and foster innovation for the bespoke modality areas.

In conclusion, digital twins represent a transformative tool for biotech and pharma companies, enabling a seamless transition towards Pharma 4.0, rewarded by gaining quality, financial benefit at less time needed for their drug products’ production.

Alix Lecomte

Upstream Process Development Scientist II

Alix Lecomte is a USP Scientist II in Process Development at KBI Biopharma in Geneva. Across her 9 years of experience in biopharmaceutical industry, she has gained strong knowledge in Cell Line Development as well as in Process Development (Novimmune, Sobi and Selexis SA). Her role within the Upstream Process Development team at KBI Biopharma is focused on leading scientific customer projects as well as participate in the digitalization journey of KBI Process Development Department.

Leveraging Bioprocess Development: Transforming Data into Business Success

This presentation will address the challenges and opportunities CDMOs face in a competitive market, having the opportunity to work with diverse molecule formats and generating extensive bioprocess data. However, the challenge is to effectively use these data to design processes that meet customer needs for scientific expertise, timelines, and transparency. To support this, innovative data recording and visualization tools were developed to streamline development and foster trust between scientists and customers. Additionally, the presentation will explore hybrid modeling methods and their potential to further leverage the generated data for business opportunities.

Dr. Amos Lu

Senior Data Scientist

Amos Lu is a Senior Data Scientist in CMC Development, Sanofi R&D. He has deep interests in mechanistic and data-driven modeling, simulation, optimization, and control of biomanufacturing processes. He is currently focused on developing and deploying these tools to accelerate and streamline upstream process development across the mammalian, microbial, and gene therapy platforms. Trained as a chemical engineer, he has a BEng from the National University of Singapore and a PhD from the Massachusetts Institute of Technology.

In-silico comparison of classical response surface and hybrid pareto optimization for feed optimization on a ground-truth dynamic metabolic model

Accelerating development timelines necessitates more efficient experiment designs and utilization of data. We first developed an in-silico ground-truth model of a media/feed optimization task with a cellular dynamic metabolic model and a batch sequential experimental format mirroring high-throughput bioreactor operations.

Using the ground-truth model, we compare classical design of experiments approach (fractional factorial, quadratic response surface) against hybrid model-based design of experiments (Latin hypercube, pareto optimization) on the metrics of maximizing ultimate titer and minimizing the required number of experiments.

Lastly, we implement realistic uncertainties for analytical variability and contamination risks and evaluate their effects of experimental design performance.

Dr. Vishwanathgouda Maralingannavar

Associate Principal Scientist

Vishwanathgouda, known as Vish, is a Scientist at Lonza, leading the Bioreactor and High Throughput group. His primary focus is on establishing small-scale culture platforms and optimizing processes in large-scale bioreactors. With a PhD in Life Science, Vish brings a wealth of research experience in mammalian cell culture. His expertise spans both academic and industrial settings, with a particular focus on media and feed formulation, clone screening, and process development.

He excels in creating experimental models and intends to utilize them for predictive process modeling to make process development simpler. Vish is passionate about leveraging his technical skills and research background to drive innovation and solve complex challenges in the biotech industry. His goal is to collaborate with like-minded professionals to develop cutting-edge solutions and make a meaningful impact.

Application of a miniaturised sub-ML cell-culture system to mimic culture performance in lab-scale bioreactors

As the field of upstream bioprocessing continues to evolve, there is a growing need for methodologies that are both efficient and data driven. High-throughput miniaturized cell culture systems have emerged as powerful tools in the pursuit of generating large datasets for precise process modelling. This study focuses on one such system that we have developed to accelerate data acquisition and enhance our understanding of perfusion processes.

Commercially available high throughput systems are limited by high operating costs and limited options for customization. We use 96-deep well plates (96-DWP) to overcome these challenges and achieve four orders of magnitude miniaturization. We have adapted to discrete perfusion method to 96-DWP to mimic a perfusion process stage. The experimental model was tested on multiples clones across products and expression systems. The model could predict cell culture performance in lab-scale bioreactors. Cell concentrations reaching ~50×106 cells/mL were achieved in 96-DWP. We estimate this model reduces operating cost to approximately 1/10th and reduces the process duration by 3 days, while increasing the number of parallel cultures multifold.

Thus, a high-fidelity model that mimics culture performance of a lab-scale bioreactor was established. We have studied its applications in bridging the gap of screening tools and validated scale-down models established in lab-scale bioreactors. Such a system allows us to gather process knowledge in early stages of process development such as clone screening allowing us to shorten process development timelines. This work highlights the significance of high-throughput miniaturized cell culture systems in generating comprehensive datasets for process modelling.

Dr. Ben Stevens

Director CMC Policy and Advocacy

Ben Stevens is a Director of CMC Policy and Advocacy at GSK and has nearly 15 years of drug discovery and regulatory experience. Prior to GSK, Ben was a Director of Regulatory Affairs CMC at Alnylam where he led the clinical regulatory CMC development of vutrisiran prefilled syringe in over 30 countries, and the initial US NDA and EU MAA submissions, including one of the earliest Notified Body assessments under the newly implemented EU MDR.

Before Ben joined Alnylam, he was a Principal Consultant at PAREXEL and an acting Branch Chief in the Office of New Drug Products (ONDP) at the FDA. At the FDA, Ben worked closely with several key policy groups (OPPQ, ORP), partnered with CDRH on matters related to combination product review, and was a government liaison to USP. Before FDA, Ben spent seven years in medicinal chemistry R&D at Pfizer and Merck. Ben has broad regulatory CMC experience in small molecules, peptides, oligonucleotides, botanicals, and combination products.

At GSK, Ben leads CMC policy and advocacy for several priority areas, including biologics, CGT, oligonucleotides, and advanced manufacturing. Ben represents GSK in numerous external trade and association working groups (e.g., PhRMA, BIO, IQ, EFPIA, NIIMBL, ISPE, PDA, Biophorum), where he has led and supported policy positions and interactions with global regulators. He received a Ph. D. in Chemistry from the University of Pittsburgh, a M.P.H. from the Johns Hopkins Bloomberg School of Public Health, and is a co-author of over 30 publications and patents.

Regulatory Considerations for Pharmaceutical Manufacturing Process Models

The integration of process models into pharmaceutical manufacturing has gained considerable attention in recent years due to their potential to enhance efficiency, productivity, and quality control. With the growing interest in the application these models to regulated products and processes, regulators and industry alike have weighed the risks and benefits of this technology and begun to formulate general approaches to ensure their safe and effective use. Early initiatives, such as FDA’s FRAME and the EMA Quality Innovation Group (QIG) Listen and Learn Focus Groups (LLFG) have begun to collect information from industry stakeholders and define specific CMC considerations, although guidance is limited at this time. A recent QIG concept paper has been instrumental in offering insight into the current thinking of EMA regulators.

This presentation will highlight some recent GSK case studies that rely process models, providing context within the evolving regulatory landscape. An industry perspective of the perceived hurdles for broader implementation and use will be presented. Existing guidance and technical definitions will be presented. The discussion will address critical elements these models such as model risk (impact), performance, validation, regulatory registration and lifecycle management, and aspects of GMP compliance. We will share feedback and experience from recent regulatory interactions. Next steps will be proposed and used to catalyze further audience engagement and feedback.