I am working on a 510(k), which would introduce some machine learning models into the data processing pipeline of our software (which is part of a medical device, not SaMD).
We did have a pre-sub meeting with the FDA, and they seemed to want a huge amount of details related to every possible aspect of the data collection and algorithm development. I am trying to write up the description now.
I was wondering if anyone has successfully submitted a device that includes machine learning models and how it went. Is there any advice I could use to hopefully not get a million questions back from them after submission.
Also, on a curiosity point, we opted not to submit with a Pre-Specified Change Control Plan, as that would have required us to use the De Novo pathways.
Thanks for the help!