Predicting wrist kinematics from motor unit discharge timings for the control of active prostheses.
Tamás Kapelner, Ivan Vujaklija, Ning Jiang and 4 others
PMID 30953528WHAT IT FOUND
Predicting wrist movement from decomposed motor unit activity was more accurate than predicting it from conventional EMG features in normally-limbed adults and a man with a transradial amputation.
This is offline prediction evidence, not a tested prosthesis outcome.
Key findings
01Motor unit spike train features predicted wrist angles more accurately than conventional time-domain EMG features in all subjects included in the statistical analysis and at all tested movement speeds, with average R² scores of 0.77 and 0.70.
02The proposed model-based neural method outperformed the other decomposed motor unit feature sets in most comparisons and never underperformed them significantly.
03For the participant with a transradial amputation, the neural method raised the offline R² score from 0.52 to 0.64, but this participant was not included in the statistical tests.
STILL TO COME
How it was doneWhat they found
Read the rest of this summary
You get three full summaries a month, free, and we do not ask for a card. Search, the TL;DRs and your library stay unlimited either way.
What it does not show
The study used offline EMG decomposition and prediction, not an online prosthesis controller, so it does not show how a patient would use a device in real time. Improvements in offline control scores did not demonstrate improvements in clinical scores. The statistical analysis included only the normally-limbed participants; the participant with a transradial amputation was reported descriptively. The sample was small: five normally-limbed men and one woman, plus a man with a transradial amputation. EMG decomposition was imperfect and could miss or misidentify spike trains, especially during movement. Only single wrist movement directions were tested, not simultaneous control of multiple directions. The estimators were tested only on tasks they had been trained on. The effect of movement speed on performance varied by subject.
The easy way to misread this
Do not conclude that this neural method will improve a patient's real prosthesis control. The study tested offline signal prediction only, did not run an online controller, and included five normally-limbed men and one woman plus a man with a transradial amputation.