Can you elaborate a little on the "right way?" If you're just looking for a jerk, a threshold ought to do (
you could also try my boss).
If you're worried about getting a noise jerk accidentally (which from my testing doesn't seem to be much of a problem), I've been working on a manager for binning together groups of signals, but that's mainly because I'm outputting a LOT of data to files, and I don't really need 125 Hz for my application.
As far as signal processing, you can always do stuff like take the peak height relative to recent averages, and find stuff like the full width at half maximum, and that'll give you a pretty good idea what the pulse looked like. I'm not really sure what I'd do with it, but it'd be an interesting puzzle to work at.
[ June 10, 2008: Message edited by: Nick George ]