Reflecting on our discussion during the meeting, here are my thoughts:
1. Type of feedback (auditory or vibratory):
Since there is no established literature as to what type of feedback is better,
will it be possible to record data in which some kids are given audio feedback
and a few other kids that are given vibratory feedback (in case this feedback
modality is going to be considered). In that way, we can avoid the issue of
learning if the same kids wherein they will end of doing more trails when the
same kids are trained using both types of feedback. By doing few kids with
audio feedback and few other kids with vibratory feedback, it will provide you
with a means of comparing which one is better.
2. Follow-up data recording
It is a must for a project of this nature. But it may be too premature to do it
right now. The first thing (step-1) to do is to establish 'proof-of-concept'
(this was the thought I had when I mentioned about feasibility testing) i.e.
demonstrate that the concept will perform comparable or better than the current
practice - i.e. it measures the same variables or more that give a better
quantification of assessing the training (outcome measures) given with the
WiiCane and that it provides more detailed information pertinent to training
aspects e.g. allows more detailed measurements (what measurements), provides a
training report at the end of each walking trial (### deviations or veering,
initial posture, final posture, time taken to finish, number of times feedback
was given, difference between the initial and final measurements to reflect +ve
ot -ve change, etc)
This step will help to establish what parameters need to be measured and is the
WiiCane a more descriptive and informative training modality as compared to the
normal cane training.
Once the proof-of-concept is established then it will be easier to build a case
to show that it provides better training. Then, step-2 will be to do record
data in follow-up sessions (some kids with WiiCane and some kids with the
normal training) and seeing how long it takes to learn the skill and how well
do they learn the skill etc.
For now, these are my thoughts.
Subscribe to:
Post Comments (Atom)
3 comments:
Archie, thanks very much for this excellent discussion about how we should structure the project to maximize the quality of our experimental results. As I mentioned, we do have a feasibility stage as you describe (what we are calling a verification experiment) in which we will use Rob's optical apparatus as a benchmark for comparing the accuracy of our measurements. the key here will be to identify gestures that are trackable with our wiiCane apparatus, that also meet our requirements. First, each parameter that we choose must also be measurable by Rob's system, and second, each parameter must represent one of the skills we have identified as coming together in the skillful cane traveler using classic two-point touch technique. These are: horizontal arc sweep slightly wider than the person's body, hand centered on the bodies vertical midline, wrist isolation, and gait/sweep synchronization. One more characteristic of two point touch that we have been discussing is that the vertical arc the cane tip describes with each sweep should be very shallow. Of these, we think that we will not be able to measure cane-gait synchronization (and there is some disagreement in the literature over how important this is), but we are pretty sure that we can measure the other things. What we have to do now is to make sure that our algorithms work for converting the raw accelerometer data into usable values. We are having a planning meeting on Wednesday where we will address this question, and after that I will be able to offer some more details. but you are exactly right to say that we won't know if our system is accurate unless we have something to compare it against (which I think we do, but we shall see...)
Archie,
Thanks for the well-articulate remarks. I have two contributions. First, I would not totally agree that we do not know what feedback works. We have two kinds of evidence. We have Guth using spoken language (which is of course auditory) with veering, and empirical quantitative evidence showing this. I think clearly this works. The second evidence we have is over 60 years of mobility teachers speaking to their students (or signing), which also seems to work. It is true that we don’t have anything on non-verbal auditory or tactile reinforcers for mobility-related tasks, and that is the question --- is there output that can be produced readily by the device that can increased desired behaviors? The only primary reinforcers we know for sure are food and praise in a verbal (sign or spoken) modality. I would like to know from others why we wouldn’t want to use properly intoned human speech feedback – is there an argument against this that I am not seeing? The added equipment would be inexpensive off-the-shelf electronics.
Second, Steve, I don’t think we have fully addressed Archie’s real question. Feasibility and ‘proof of concept’ are not only benchmarking and comparing our output with Rob’s more-sophisticated device, because the WiiCane is a teaching device. It is not primarily or only a recording device. Archie, you said, “demonstrate that the concept will perform comparable or better than the current practice . . .” If subjects come to trials with a base-line of skills, isn’t that the result of current practices (traditional O&M instruction in a given quantity), and if feasibility trials shows changes in behavior in the positive direction, isn’t that a proof of concept?
Or have I misunderstood what you think should be the early feasibility steps we should take?
Gene,
Concerning your question: If subjects come to trials with a base-line of skills, isn’t that the result of current practices (traditional O&M instruction in a given quantity), and if feasibility trials shows changes in behavior in the positive direction, isn’t that a proof of concept?
Yes... Are there any measures currently being recorded to characterize base-line skills? If so, the same measures will need to be recorded with the WiiCane training. Then positive results will help to establish proof-of-concept.
Post a Comment