Speaker
Description
The development of drugs/vaccines is very expensive and comes with huge failure rates, financial risks. For companies to be successful portfolios need to balance financial risks, investments and potential revenues. Recent years have seen big pharmaceutical companies focusing on specific diseases or conditions, potentially as a mean to understand the financial risks and returns better. In this context the role of process development is to deliver robust processes in short timelines and at limited technical risk. The Quality by Design (QbD) paradigm is seen as a measure to keep technical risks in check, as it helps to systematically explore the process knowledge and improve the understanding. However, QbD guided bioprocess development is time effective only if knowledge is transferred from one project to the next (horizontal knowledge transfer), one scale to the other (vertical knowledge transfer). Today knowledge is transferred across scales and projects when assessing the technical risks. However, this form of knowledge transfer is limited and despite the wide spread of platform processes, to some degree the process needs to be developed de novo for every new drug/vaccine candidate. This leaves a huge potential to accelerate process development.
In this contribution, we show how advanced machine-learning and hybrid modeling approaches can be exploited to transfer knowledge between scales and projects. In particular, we present a novel embedding approach that allows for transversal data analysis across process runs with different cell-lines and products. We showcase the added value of the approach for an upstream bioprocess development case, concluding that process development could be more efficient and multiple times faster.
| Speaker Country | Switzerland |
|---|