[time-nuts] Off topic project sort of heart rate monitor NEED BEATS PRE MINUTE TO ANALOG VOLTAGE

[Jim Lux]

On 2/19/13 10:58 PM, Hendrik Dietrich wrote:
> Good morning Paul,
>
> First: I almost spilled my coffe as you bear a similar last name as a
> guy at a national research lab who does exactly such things :)
>
> Even when it makes your ECG preamplifier free for other things, I advise
> you to use another sensor and put a lot of decoding and interpretation
> to your brain.
>
> ECG is not easy from a safety standpoint, as it connects to the body
> immediately. The Signal is quite complex as it is composed of not a
> single but a "burst" of waves during excitation and contains a lot of
> moving and electrode artifacts, I wouldn't try to count BPM except AD
> conversion and doing a lot of filtering and preprocessing in software.

Of course pretty much every elliptical machine, treadmill, and other 
exercise equipment has some form of ECG based measuring system, 
typically working off the handgrips.

However, I'll agree that processing the signal is troublesome because of 
motion artifacts and the like.  I've not seen a lot of algorithms for 
this, but the ones I've seen look a lot like NTP kind of filtering: you 
have an estimate of when you expect to see the next heart beat, so you 
time gate your signal to reject signals that occur too soon, or are 
missed.  Missed beats just don't update the rate, and you use the beat 
to beat timing, inverted, to update the heart rate estimate, using some 
sort of exponential smoother (probably for implementation simplicity.. 
y(n+1) = 0.75*y(n) + 0.25*x(n) is something that a 8 bit micro can do 
pretty easily)




>
> I would use photoplethysmography instead of ECG, means you shine a
> little light from a LED thru your finger or ear to a photodiode.
> Actually a optocoupler with a bit of you inbetween !
> The light at the photodiode will fade up and down with the local
> relative blood pressure in the tissue between the emitter and receiver,
> and so does the photodiode current. The resulting signal contains way
> less movement artifacts and less high-frequency parts.

You can also do it with reflection.  LED shines into your finger, and 
photodiode/transistor next to it looks the light coming back.  You can 
use an off the shelf reflective sensor as used for things like limit 
switches.

> Hendrik
>
> P.S.1.: My first post at the time-nuts
> P.S.2.: Don't let the FDA see anything :)



I've always wondered about how the exercise machines get around the 
"medical device" rules..