edm_msu wrote:I still don't get it. I'll exaggerate. Lets say that the Zeo is hooked up to someone who had an absolutely perfect night of sleep. The OUTPUT of Zeo showed a completelty horrible night of sleep. This completely unaccurate INPUT would be analyzed. As a result, the sleep would need to be "improved" since it was horrible. After many nights of trial and error, the Zeo is hooked up to the same person who had an absolutely horrible night of sleep. The OUTPUT of Zeo showed a completelty perfect night of sleep. The end result is that the repeatable and unaccurate Zeo OUTPUT caused the sleep quality to become much worse.
Am I missing something here?
Since in your hypothetical case you know that the persons had horrible and perfect nights sleep, the analytics would reverse the Zeo's output when using it as an input to the analytics. Then, since I require that to be repeatable, if the Zeo again showed a horrid night sleep, the analytics would know that means a good night's sleep and vis versa. If I used the Zeo brain waves as an input, it does not matter if Delta is 30% or 20%, just as long as it is consistent, right or wrong. If the Sleep Score was 99 when in fact it should have been 88, it does not matter as long as it is consistent when I'm using it as an input to analytics.
In commercial applications, pressures, temperatures, flow rates, etc. in manufacturing do not have to be correct in order to predict product quality, just consistent. They can even be very noisy (somewhat inconsistent). The analytics filters out the noise, and maps those conditions to product quality, even reversing their effect if necessary.