Last week our partner Ulrich Köllisch participated in the FDA/PQRI Workshop on the Regulatory Framework for the Utilization of Artificial Intelligence in Pharmaceutical Manufacturing. In this interactive event regulators and industry shared their thoughts on implementing AI/ML solutions in the GMP environment.
Here are the top 7 take home points from Ulrich:
- Definitions/glossary are a mess in the GMP-public. Here a need for clarification was clearly identified.
- The compliant use of dynamic models is not yet set, most solutions presented at the workshop lock the model and train it in defined intervals.
- Interpretability is key: ML solutions need to be documented/presented/validated in a way that enables quality units and inspectors to understand the concepts and the risks.
- Training of Quality Units (and Regulators) is one key to success now.
- You need to have a framework around ML and Data governance in place; one speaker called it “good digital practices”.
- The responsibility stays always with the human!
- Make the model as simple as possible but not simpler (following Einstein’s principle of simplicity)
GxP-CC is very happy to bring in our strength in the above-mentioned points 3,4 and 5. Feel free to reach out to us anytime!