User training for pattern-recognition based myoelectric prostheses using a serious game

BACKGROUND
Individuals with upper-limb deficiency who are fitted
with a prosthesis are normally trained in the use of
such device. During training pattern-recognition based
control, the way training improves control skills and
what kind of feedback facilitates skill acquisition is
still poorly understood. Moreover no training protocols
for pattern-recognition control have been described in
the literature that do not require an expert trainer which
limits the clinical applicability of pattern-recognition
controlled prosthetics.
AIM
The aim of this study is to investigate how different
kinds of feedback improve the control skills of patternrecognition
control.
METHOD
Nine able-bodied volunteers participated in an
initial experiment in which they either performed
conventional system training or gamified system
training with no prior user training. Based on the results
of the initial experiment, a follow-up experiment will
be conducted where four different kinds of training
regimes are investigated namely; conventional training
with knowledge of results but without coaching,
conventional training with knowledge of results and
knowledge of performance, serious game training or
training with no feedback (control). Each group will
include 15 participants, who train one hour per day for
five consecutive days. Offline and online accuracy as
well as the repeatability index, separability index and
mean semi-principal axis [1] will be reported for each
group.
RESULTS
In the initial experiment control performance was
measured using the motion test [2] and a questionnaire
was used to measure the participants experience of
the system training and evaluation. No significant
differences were found between the groups.
Preliminary results of the follow-up experiment will
also be reported.
DISCUSSION & CONCLUSION
The finding that no differences were found between
the groups in the initial experiment could be ascribed
to the way the game targets were set in the gamified
system training and to an unintuitive mapping
between movement and game control. In the followup
experiment that will be performed a fixed mapping
inspired by Radhakrishnan et. al [3] is used where the
electromyogram measured at each electrode located on
the forearm is mapped to a direction.
REFERENCES
1. Bunderson NE et. al, IEEE T Neur Sys Reh. 2012
2. Kuiken TA et. al, J Am Med Assoc. 2009
3. Radhakrisnan S. et. al, J Neurophysiol. 2008