Activities of daily living with bionic arm improved by combination training and latching filter in prosthesis control comparison.
Michael D Paskett, Mark R Brinton, Taylor C Hansen and 4 others
PMID 33632237WHAT IT FOUND
Training prosthetic control algorithms on simultaneous movements improved speed and accuracy compared to single-movement training.
Adding a nonlinear smoothing filter reduced errors but slowed performance. Users preferred algorithms that felt easier to use, even when objective task times were similar.
Key findings
01Training algorithms with combination movements (simultaneous finger and wrist actions) significantly reduced dropped clothespins and transfer times compared to training with individual movements.
02Applying a nonlinear latching filter reduced dropped clothespins and subjective workload for some algorithms but significantly increased transfer times for others.
03In complex daily tasks, objective performance differences between algorithms were small, but users significantly preferred the Convolutional Neural Network (CNN) and modified Kalman Filter (mKF) over the Multi-Layer Perceptron (MLP).
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
Study used non-amputees with a bypass socket, which may not fully replicate the sensory and mechanical experience of amputees. Tasks were easier than anticipated, causing ceiling effects that masked performance differences. Small sample size (10 and 9 participants) limits generalizability. Compensatory body movements were not directly measured or restricted.
Declared interests
Funded by Defense Advanced Research Projects Agency and National Center for Advancing Translational Sciences.
The easy way to misread this
Do not assume the preferred algorithms (CNN/mKF) are objectively superior in speed or accuracy, as performance metrics showed no significant differences. The preference was driven by subjective ease of use and lower workload, which are distinct from functional task completion rates.