Dismiss Notice
Join Physics Forums Today!
The friendliest, high quality science and math community on the planet! Everyone who loves science is here!

Data/Waveform comparison

  1. Oct 11, 2012 #1
    I am a graduate Physical Therapy student trying to complete a group, final-year capstone project. I need some guidance with what I believe to be an engineering analysis issue with our collected data. I apologize if this isn't an appropriate question for this forum.

    We are using an accelerometer to collect angle measurements on the movement arm of an exercise machine. In our project, to establish the validity of our device, we measured the movement arm of an expensive, "gold standard" physical therapy rehabilitation machine. For those test runs we have our own data as well as the data exported directly from the operator console of the rehab machine. I'd like to compare those data sets and arrive at some measurement to express the degree to which our measurements match the reference measurements.

    This is what we've done so far. We've plotted the two data sets on a common graph and aside from some occasional "spikey" noise on our data, the rendered waveforms are nearly identical. Since we cannot start both devices at the same instant, our recording begins before the reference device. To overlay the waveforms we have to "move" our data along the time axis to match the time zero of the reference by eye.

    This is where our analysis may begin to run off the tracks. The reference data is sampled every .01 seconds and our data is sampled approximately every .05 seconds. (I believe the reference data is probably well filtered and the time increments are reported exactly every .01 seconds). We applied a spline fit to our data so that our new data points occurred at the same intervals as the reference data and ran a Pearson correlation on the two sets. The correlation was 0.9988.

    From an engineering standpoint, is this a valid way to compare these waveforms? Is there a better way to find the new zero point for our data than eyeballing the plotted points? The movements we're monitoring have a period of about .5 Hz and the occasional noise spikes (caused by backlash in the arm's coupling to the dynomometer, I believe) are about 20 times faster. Since the spline fit reduces the magnitude of the noise, would that render our comparison invalid? Can anyone point us in the right direction to make this comparison properly? Sorry for the long post. Thanks in advance for any help.
  2. jcsd
Share this great discussion with others via Reddit, Google+, Twitter, or Facebook

Can you offer guidance or do you also need help?
Draft saved Draft deleted