No matter if in development or in manufacturing, biopharmaceutical companies swim in data. However, data is not analyzed due to multiple reasons: Missing availablility; Missing contextualization; Different frequency; Different dimensionality. As a result, experiments are not based on previous knowledge, creating an unnecessary waste of resources and costs. This contribution shows how to automatically organize and analyze data at the speed of its generation. We show good practices on how to achieve good data quality for achieving significant data analysis results.

Case Study presentation to explore the relationship between scientific user experience and high quality data set generation in the context of evolving scientific methods and digital transformation. Digital product innovation guiding principles, a model for informatics product team/ scientific user partnership and a summary of lessons learned will be shared.

Tech transfer projects have always been stigmatized with tight timelines and limited resources. This presentation would like to tackle this challenge of increasing effectiveness and efficiency of tech transfer challenges by considering lessons learned and what can be done better through digitalization.

We discuss the strategy of implementing process modeling for production of complex biotherapeutics and how models powered by ML can be useful to drive data value continuum along the CMC development cycle. Case studies will be presented on the use of hybrid, mechanistic, and data-driven models for upstream and downstream, including potential applications of a novel physics-informed machine learning algorithm.

In the slowly evolving landscape of bioprocess development and manufacturing, digital bioprocess-twins have emerged as potential accelerators. While advanced algorithms are at the heart of this endeavor, they are just one piece of the puzzle. This presentation will illuminate the essential stages in developing robust process models, encompassing experimental design, customized modeling strategies, smooth scale-up processes, and the real-time application of models for effective monitoring and control. Concrete examples from both upstream and downstream processes will be provided to enhance comprehension of these principles.

As companies in the pharmaceutical manufacturing industry undergo digital transformations, they face numerous questions around the transformation. What are the goals? What are the benefits?  What does good look like? In this presentation, we examine digital transformation through the lens of digital plant maturity models, which help companies understand the current state of their plants and prioritize their digital investments as they progress up the maturity curve. We present use cases and examples on Digital Twins, In-Line Monitoring/Real-Time Release, and Model Predictive Control. 

Ex-vivo cultivation of iPSCs for the production of red blood cells is a promising therapeutic alternative to donor-based cell transfusion, yet scale-up of this bioprocess remains challenging. In this work, highly-resolved large-eddy simulations were performed to compare the hydrodynamics of a 125mL shake flask and 250mL bioreactor for various operating conditions, to help translate shake flask operating conditions to bioreactor operating conditions, which ultimately contributes to faster process development times.

In this talk, we discuss model predictive control (MPC) strategies for CHO fed-batch cell culture. MPC strategies have been investigated for their potential to achieve consistent end-of-run titer in the presence of unexpected process disturbances (i.e., iVCC deviations, pH or temperature controller errors). First, we discuss two different modeling frameworks and their integration with MPC. Then, we compare the impact of process disturbances on the process with and without MPC.

Mammalian cell lines require complex media. Cells utilize the nutrients available to them as building blocks for biosynthesis, but also as substrates from which they derive the energy to drive biosynthesis and cell maintenance. Here we show how the rates of catabolic reactions can be dissected from the rest of metabolism, and show that catabolism governs the rates of growth and product and byproduct formation in antigen-producing CHO cells.

Screening for CMC protein liabilities is a crucial, although costly, step in mAb drug product development. The diffusion interaction parameter, kD, has been demonstrated to be a highly useful predictor for CMC liabilities, and kD, in conjunction with B22, can be predicted using molecular-scale simulations. After parameterizing coarse-grained simulations, we have determined that these simulations are both robust and efficient for predicting self-interactions of monospecific, commercial mAbs.

• Factors to be considered when deciding on building an in-house software solution vs. off-the-shelf offerings
• Tools to build a successful digital ecosystem—from specific code packages to ELN and data exploration platforms of choice
• Infrastructure supporting the functions from Data Engineering like data capture, data storage, and ETL pipelines; to Data Science like model versioning, data exploration, and advanced analytics
• Challenges—assay nomenclature harmonization, standardization of data capture and data processing workflows, and strategies to routinely generate FAIR data
  

Worst-case studies are a facet of late-stage process characterization that are used to model the combination of factors that have the least desirable outcome in a manufacturing process. Both screening and response surface DoEs were executed to identify the highest risk factors on a 14-day fed-batch cell culture process. These models were then leveraged for designating the Upstream Control Strategy AORs prior to PPQ manufacturing.

This work introduces a groundbreaking digital twin designed to enhance the production of recombinant-adeno-associated virus (rAAV) within baculovirus/Sf9 cultures—a platform responsible for producing 50% of commercial rAAV-based gene therapies. A mechanistic model is developed to systematically identify bottlenecks within the intracellular pathway for full rAAV capsid formation in producer cells. After experimental validation, the digital twin indicates genetic modifications and process enhancements aimed at boosting overall platform productivity.