R-R detection

I am seeking a software program that detects heart rate variability in
time-series signals. Does anyone know where I could find such software
as freeware or shareware? Or some helps to program myself.

 

This sounds like *way* too specialized to be something offered as
freeware or shareware, or even as a consumer product. There may
be software that a research lab could provide, but you'd have to
figure out which lab to contact. Perhaps a literature search to see
who is doing research in this area?

Assuming you already have the time-series recorded as a file,
you can try having a go at this yourself. However, your results
will depend entirely on how good a signal you have. Your EKG
signal must be really clean and stable to pick off peaks reliably.
That means the subject doesn't move around during recording.
Or, if you are only looking for gross interval variability, you might
get better results with an optical flow sensor. (Haven't used
those myself.)

If you have a clean-enough recording, you may be able to find
some threshold level that will reliably detect a QRS complex
or flow peak. Then you simply note all the distances (in samples
or msec) between these events and sort them to a histogram
that shows all the intervals. If the variability is low, they will
tend to cluster in a central peak at the average interval.

However, a simple threshold may not work well, and you will
need to have something fancier. One method is to have a
series of "hoops" the signal must "jump through": It must
be below a certain level to indicate the inter-event time,
then it must rise at a certain rate to indicate the start of
the event, etc, etc. These are typically tweaked by hand
for best reults. Yu can also try a simpler approach by
having the threshold self-adjust to some percentage of
the last peak seen.

Good luck!


Bob Masta
dqatechATdaqartaDOTcom

D A Q A R T A
Data AcQuisition And Real-Time Analysis
www.daqarta.com

 

I worked with a box that did exactly what the Frankis seems to be
talking about. The peak detector was written by some other guy, but
it was really simple - just a couple of filters implemented in
software. It gave a very clean, rock-stable one-pulse-per-beat
output, which then got counted.

I don't recall how he wrote the filters, and didn't really
understand it anyway, but basically, take the transfer function
of whatever filter you're building, and code it. Catching a
QRS pulse is almost trivial, if you know what you're doing.
At least Monty Barker (that was the guy's real name, I swear)
made it look easy.

I suppose you'd make something with a response peak at the
fundamental of the (approx) half-sine at the start of the wave,
and even a little analysis of the waveform itself. (this box
even spotted V-Fib - it's just another filter.)