Due to the large number of regulations, the pharmaceutical industry has remained in the Champions League for machine vision. Accordingly, the topic of AI seemed a long way off (due to the even more difficult validation, among other things).
Nevertheless, together with Takeda, we dared to try and perform an AI retrofit on one of their AVI systems. Here is the result at a glance:
And here are the facts about it in a nutshell:
✅ Integration of our AI box into the AVI environment.
✅ Improvement of the performance of the long-calibrated system in less than three months.
✅ Reduction of the False Reject Rate (FRR) by approx. 24% and simultaneous increase of the Detection Rate (DR) by 3.8%.
✅ Establishment of an approach for continuous improvement of AI software in pharma.
You can read about this in the Nov/ Dec issue of ISPE's Pharma Engineering magazine. Here is the link to it: https://ispe.org/pharmaceutical-engineering/november-december-2022/ai-retrofit-tackling-challenges-cutting-edge