2020 commercialisation, page-340

I think for all the auditing purposes that go with developing a diagnosis model it must be housed on a central platform and accessed through an API. This is also the only way to do it and charge on a per use basis. If it isn't connected to the bet charging per use is next to impossible. Ive experienced this personally.
What you're saying Victa wouldn't be for such a machine learning model. I know these models. They run on Sagemaker (Amazon), h2O, tensorflow. That all run on the cloud and report back results.
i very much doubt the would be housing the model on the phone. And this is a real issue in 3rd world countries. But even 3rd world countries in hospitals have enough bandwidth to support a couple of kilobyte upload for a sound file so I don't see it as that much of an issue.
from all communications from management, the issue is due to the difference in microphones across devices and what preprocessing they have to do around this. Having their own ruggedized device alleviates this problem as it's a controlled unit whereas bringing it out for Android as a whole introduces a lot of variability.
imo