New Research: PUSH Validity

Measuring and monitoring lift velocities is something that has been rapidly growing in popularity- for good reason. We can get all kinds of useful information from paying close attention to the bar velocities our athletes are able to create. We can use this information to direct training, to get a feel for athletes’ progress, and to assess preparedness. There are quite a few devices out there that do this. Tendo and GymAware are probably the most popular, but there are other companies, like PUSH and Assess2Perform that are coming up with alternative devices that don’t require wires. This is a huge advantage over wired devices, since you can use more exercises and you’re not necessarily tied (pun intended!) to any particular station. While I haven’t used the Bar Sensei, I do have a bit of experience with PUSH.

In the fall of last year, we decided to assess some of the metrics coming from the PUSH unit. Both Dr. Sato and I could see the usefulness of the device, and we both recognize that practictioners are interested in using technology like this. However, there is very little information about the efficacy of these newer devices. We set out to evaluate, albeit in limited fashion, some of the metrics reported by the units.

In our paper, we evaluated dumbell overhead presses and dumbell curls- the idea was to check out a relatively linear exercise and a curvilinear exercise to make a general inference about other exercises with the same characteristics. We evaluated two variables that the device kicks out- average and peak velocity. Since the PUSH device is making an estimation of how the barbell is moving, we compared their metrics to those same metrics calculated from hand markers in a 3D motion capture setup. These markers are really close to the position of the bar (the hand is holding the bar, so the bar & hand marker positions are going to be very similar). We compared left side to right side values from both PUSH and motion capture, and upon finding they were similar side-to-side, we pooled the left and right side data. Overall, we found that the PUSH and motion capture values were similar, and correlated strongly. There was however a systematic bias- meaning that there was a somewhat consistent difference between devices for measurements taken. Not a big issue necessarily, since we won’t be replacing motion capture with the PUSH unit anytime soon. This just indicates that if you’re going to use one type of device, stick with it, since the values between various devices may differ. We also discovered that the PUSH device had some issues with outliers- there were a few cases of weird underestimations of velocity in the DB press, which you can see in the scatterplots of the paper. If you’re using PUSH, just watch for weird values, and make a judgement call as to whether the numbers make sense or not.

The moral of the story is that PUSH seems to be pretty good, at least for the exercises tested. It had some minor issues, those outliers I mentioned, but overall, not bad for the price point. The study we did does not completely validate the device however, and I hope folks won’t assume that this study proves that everything about the PUSH device is solid. I don’t mean to imply that it isn’t good, rather there is more work to be done. We need to test out other exercises, we need other research groups besides us to evaluate it, and we need to test it out against their competition. Most notably, we need to check it against the big players- Tendo and GymAware, and against the other upstarts, like Bar Sensei (Assess2Perform). However, to our knowledge, this is the first study to look at the units, and so far, they look pretty good.

Check out our study here:

Validity of a Wireless device measuring velocity of resistance exercises

PS: In the interest of full disclosure, I should mention that we were given the devices by PUSH for the study. I also wrote a couple of blog posts for their website. The study itself was independent of the company, and we have no financial stake in PUSH.