The accelorometers can give you the characteristic of motion....problem is that the correlation to biology will be subjective since while the mechanics are reproducible, what's inside every head is different
You could record endless data from the contactor, contactee......you would have everything in terms of the information about the physical nature....but accuracy will be compromised due to little things like helmet fit....will every helmet be calibrated to its head....will it change if the guy gets a haircut? How will the effect of the jaw slamming shut be included into that data? Constant base lining through periodic brain scanning could help in an individual basis
Awesome modeling challenge, but I think we all know the ultimate answer is actually very simple......contact will result in concussions
tl;dr version:
With all respect, you are thinking too small.
Data would be useful not just for concussions, but for the study of the relations between acceleration and many other neuronal, cognitive, and behavioral factors, in addition to the efficacy of different types of safety equipment in real-world conditions. Concussion is the neurobehavioral tip of the iceberg: focusing only on that would be like oncologists focusing only on people with stage 4 cancer. There's a lot more to it (sub-concussive data).
Long version
Regardless of the practical issues of implementing it, the claim that such data would not be useful seems simply patently false. Of *course* there will be a correlation, but what we are talking about is quantitative data that will allow us to go deeper. Science starts with correlations, but then delves in for causes and mechanisms. For instance, we know there is a correlation between smoking and cancer, so should we not study it in more detail, look at smoking frequency and relation to cancer, but also lung tissue damage, precursors to cancer, etc., try to get at the mechanistic details that lead up to it? This is where the scientific and clinical questions are often the most interesting. Not in the coarsest clinical expressions at the extremes, but the incremental changes that lead up to putative syndromes such as CTE. This is where intervention is still possible.
Data from humans could suggest useful hypotheses that could be tested in animal models, in which useful clinical solutions could be tested before being sent into the pipeline for practial applications in people. Or we could find that certain helmets and other protective technology clearly perform better than others in practical application (versus the laboratory of manufacturers). This would be a huge benefit to the players.
The suggestion that we would not want to have access to one of the key pieces of information relevant to a major problem in professional sports is impossible to justify on theoretical grounds. If it were too costly, that is one issue, but the bald claim that even if we could get such data, it would not be desirable? Nuts.
Obviously acceleration is just one piece of data, but an extremely important one. If the NFL really wanted to attack this, it would be one of the more obvious and useful data streams to provide neuroscientists and neurologists.
As far as practical challenges getting the thing to give accurate readout, that is a different issue, and if that were to make it impossible, then fine then it would not be possible. However, if haircuts are going to be the excuse, then we need to hire better engineers...It would likely be embedded in silicon at some optimized place in the helmet, so things like talking and jaw moving wouldn't significantly influence it (and those are going to have very low acceleration that can be filtered out in the analysis stage).
Noise and variability are all part of data analysis, science deals with variability all the time, it doesn't mean there still aren't correlations, or you can't troubleshoot/denoise/filter the data and find patterns. That's what statistics was invented for: to deal with real-world variability in data. Crash test dummies experience huge acceleration magnitudes and deltas and the data are extremely helpful. NFL players clashing produce acceleration and jerk a few orders of magnitude larger than when they chew gum, so my hunch is that the data would be pretty damned clear.
The thing to do is try it, not to talk about it endlessly. People who crap on cool ideas often end up not implementing them. Those who implement them to see if they will work tend to make the coolest things. I've seen this so many times I don't even listen to the poop talkers anymore, I go to the people who say "Let's build it and see how it goes" rather than the engineers whose first instinct is to say "Here's why that won't work." I work with both types, and the former always give me better widgets, faster, while the latter are still diddling away in AutoCad.
NFL players provide a real-life laboratory to collect invaluable data about brain injury. Do it Roger.