In order to correlate with heath issues your going to need a standard to work from. Which makes collecting most of this info before 'fit-bit 12.0' or whatever somewhat pointless.
What exercise data can tell you is if your gaining or losing ground. So, if you do 10k in the same time but your heart rate is higher or lower that means something. But it's not going to correlate with a say a 1% increase of diabetes because we just don't run studies with that much detail yet.
The site seems to be showing that it has gathered a lot more than just eating and burning calories, safe to say such intensive data can be used to correlate with certain standard illnesses to be cautious.
Your data over a period of few months might make more sense to understand what ill-effects certain things have had on you. Think about 1 year's worth of your data being analysed to tell what certain things have had certain health related changes in you. Of course it still would take time to know how the changes you make will change things in you but based on your past data it could predict perhaps?
What exercise data can tell you is if your gaining or losing ground. So, if you do 10k in the same time but your heart rate is higher or lower that means something. But it's not going to correlate with a say a 1% increase of diabetes because we just don't run studies with that much detail yet.