A Model-Based Process Development

Discover how hybrid models optimize bioprocessing

Watch Now

Dr. Moritz von Stosch

Chief Innovation Officer
DataHow AG

Overview

Dr. Moritz von Stosch, CIO of DataHow, introduces an innovative model-based approach to process development using hybrid models. He explains the key concepts and technologies behind this methodology.

Hybrid bioprocess models, which incorporate AI and machine learning, offer significantly faster learning within bioprocess development, supporting accelerated development, more robust processes, and higher yields. This active learning approach reduces data requirements and extrapolates to multiple products and/or process features to maximize insight while minimizing experimental effort.

DataHow’s transfer learning approach allows insights from small scales to be transferred to large scales, resulting in a significant reduction of large-scale experiments and, therefore, time and cost. This approach can also be applied across products, enabling significant development efficiencies on new developments.

DataHow’s hybrid process models and transfer learning approach work in concert with digital twins, a digital representation of a physical process with a robust process model at its core, to run in silico simulations vs. wet-lab experiments, perform “what-if” scenarios, and enable process monitoring and alerts for operator intervention.

The impact of hybrid models on process development will be demonstrated through real-world examples.

Key topics will include:

  • How hybrid models accelerate process development;
  • In silico experimentation replacing a large percentage of wet-lab experiments;
  • Leveraging existing data and knowledge for faster process development;
  • How hybrid models enable insight-driven process characterization and scale-up;
  • Improved collaboration across the bioprocess enterprise.
Transforming digital
Bioprocessing