3 ms·

Computers, and the like generally don't provide much functionality that's useful... -ch4s3 in 1950

by breck 6y ago

ch4s3 6y ago
That's not at all what I'm saying. My point is that fitbit and similar wrist mounted consumer devices aren't currently useful for real health purposes. I was responding to: > Maybe, but since we don't have currently good way to use our health data for health purposes My criticism is that the is NO WAY you could use this data, because the data is junk. I even pointed out that there are larger devices that do work. But there are real physical limitations in trying to measure heart rate through the wrist by bouncing light off of the veins. Measuring blood flow through tissue downstream from the heart is an indirect measurement at best. Skin, sweat, tissue composition, and motion all add confounding variable that are hard to correct for in software. The data you do get from the device is do you regularly exercise while wearing the device or not.
- breck 6y ago
  I won't disagree that the data could be better. It can always be better. But it gets better. With every. single. update. You're just wrong about no ways to use this data. I've personally dramatically improved my health via wearables over the past 7 years. A close friend with a heart condition has perhaps had his life saved by these things. Check out the FitBit forums or subreddits. When you have a lot of data with high dimensionality and apply modern ML/DL methods you can discover connections that just aren't visible with fewer dimensions. Healthcare is going to have its "What Just Happened?" moment sometime in the late 2020's.(https://moalquraishi.wordpress.com/2018/12/09/alphafold-casp13-what-just-happened/ https://moalquraishi.wordpress.com/2018/12/09/alphafold-casp...)
  - ch4s3 6y ago
    Color me skeptical, but I've read a mountain on research on this very topic, and know some people on the medical side of this space and it just doesn't seem like the quality of the data is there for medical use. You could maybe do something with ML/DL on resting heart rate measurements for some kinds of people (thin but not too thin, and with lighter skin). I'm totally willing to believe that cool things are happening in med-tech, but I don't believe optical blood flow sensors mounted in a watch band are one of those things. If someone produces some real research where they get meaningful diagnostic info out of these devices, then I'm open to changing my mind.

Computers, and the like generally don't provide much functionality that's useful... -ch4s3 in 1950

by breck 6y ago

Computers, and the like generally don't provide much functionality that's useful... -ch4s3 in 1950

That's not at all what I'm saying. My point is that fitbit and similar wrist mounted consumer devices aren't currently useful for real health purposes. I was responding to: > Maybe, but since we don't have currently good way to use our health data for health purposes My criticism is that the is NO WAY you could use this data, because the data is junk. I even pointed out that there are larger devices that do work. But there are real physical limitations in trying to measure heart rate through the wrist by bouncing light off of the veins. Measuring blood flow through tissue downstream from the heart is an indirect measurement at best. Skin, sweat, tissue composition, and motion all add confounding variable that are hard to correct for in software. The data you do get from the device is do you regularly exercise while wearing the device or not.

breck 6y ago
I won't disagree that the data could be better. It can always be better. But it gets better. With every. single. update. You're just wrong about no ways to use this data. I've personally dramatically improved my health via wearables over the past 7 years. A close friend with a heart condition has perhaps had his life saved by these things. Check out the FitBit forums or subreddits. When you have a lot of data with high dimensionality and apply modern ML/DL methods you can discover connections that just aren't visible with fewer dimensions. Healthcare is going to have its "What Just Happened?" moment sometime in the late 2020's.(https://moalquraishi.wordpress.com/2018/12/09/alphafold-casp13-what-just-happened/ https://moalquraishi.wordpress.com/2018/12/09/alphafold-casp...)
- ch4s3 6y ago
  Color me skeptical, but I've read a mountain on research on this very topic, and know some people on the medical side of this space and it just doesn't seem like the quality of the data is there for medical use. You could maybe do something with ML/DL on resting heart rate measurements for some kinds of people (thin but not too thin, and with lighter skin). I'm totally willing to believe that cool things are happening in med-tech, but I don't believe optical blood flow sensors mounted in a watch band are one of those things. If someone produces some real research where they get meaningful diagnostic info out of these devices, then I'm open to changing my mind.