One.) At a set, low speed we all walk pretty much the same - at least when it comes to the movements the ratchet was picking up.
Two.) Might be interesting to note that they had a pretty small sample size (n=9), and some people were getting more a benefit than others. I think one participant was up to like 10%.
As the article says, most of it is fairly simple. Just a spring for a cable and a ratchet that moves with the motion of your foot (kind of auto tuned), and grabs the spring.
If the spring was remotely tensible, you could probably just walk and adjust it until it felt best.
You could probably also do that at various speeds and then create a smooth mapping of ensuing parameters for various gaits.
So my question is more about the behavior of the implementation, not about the design space for it.