What's the plan for obtaining other comparable recordings?
A single patient recording might be able to yield some insights specific to that person, but they wouldn't provide much information to generalize from even if they're over a very long period of time.
To have any power to evaluate machine learning effectively you'd need at least 5-10 subjects (IMO) with most confounding factors controlled in some way or another. At that scale it doesn't seem practical to have people self reporting as they find the database.
I think there's actually a lot to learn from a single patient recording. The ultimate aim is to give people advice in real time for them to better manage their health. This needs to be customized to each individual, so working on a single patient recording can help us figure out the method to do that (and whether something like that is even feasible or not). This would be a “proof of concept”.
But I do agree that more data from more patients would be even more interesting. To be honest, the plan to get more data is not well defined yet, but I'm going to reach out to many people through different methods, we'll see what works and what doesn't.
I'm also hoping to aggregate other kind of experiments all in one repository.
A single patient recording might be able to yield some insights specific to that person, but they wouldn't provide much information to generalize from even if they're over a very long period of time.
To have any power to evaluate machine learning effectively you'd need at least 5-10 subjects (IMO) with most confounding factors controlled in some way or another. At that scale it doesn't seem practical to have people self reporting as they find the database.