Hi,
A customer in pharma has some new tech for manufacturing peptides quickly. E.g. a peptide of 40 amino acids would previously have taken 1 week. Now it takes 8 hours.
The regulator needs to know that the process is in control.
They can collect data from about 40 different sensors (Temperatures, Powers, Pump Speeds, Temperature Set Points) at high frequencies. But they would probably collect every minute. Obviously there will be a correlation structure to this data. So multivariate techniques should be appropriate.
The question they have is if they collect data over a number of “typical”, “good” batches could they then use this data to demonstrate that the process is in control for future batches and the batches are fit for release?
As a further step they might be interested in seeing if this can be applied in something close to real-time. That is, the solution would tell them if a batch is atypical during the process so they need to consider intervening in some way.
Anybody got experience of this kind of thing? Any ideas for what a solution might look like?
My thoughts are that this should be possible. I think there are other software solutions that are more targeted for this kind of situation (the customer would like a JMP solution). Perhaps Functional Data Explorer in JMP14 Pro could be useful.
Thanks,
Phil
