Improving upper-limb prosthesis usability: Cognitive workload measures quantify task difficulty
DOI: 10.1101/2022.08.02.22278038
archive: archived pipeline: cataloged
Summary
medRxiv preprint comparing multiple cognitive workload measures during upper-limb neuroprosthesis use. Participants controlled a virtual prosthetic arm via surface EMG and performed a virtual target-control task at easy (large targets) and hard (small targets) difficulty levels. Two participant groups: one (n=1 amputee + n=10 non-amputees) measured with DRT and EEG (parietal alpha, frontal theta, P3 ERP); the other (n=1 amputee + n=10 non-amputees) measured with ECG (LF/HF HRV ratio) and pupillometry (task-evoked pupillary response). All participants completed NASA-TLX. The DRT, LF/HF HRV ratio, task-evoked pupillary response, and NASA-TLX significantly differentiated easy from hard; EEG measures did not. The DRT was easiest to deploy and most sensitive across and within subjects.
Key finding
Among objective workload indices applied to prosthesis use, the DRT was the most sensitive and easiest to deploy; LF/HF HRV ratio, task-evoked pupillary response, and NASA-TLX also discriminated task difficulty, while EEG alpha, theta, and P3 measures did not.
Methodology
Two-group within-subjects comparison. Each group performed easy vs. hard virtual target control tasks via sEMG-driven virtual prosthesis. Group 1 (n=1 amputee + n=10 non-amputees) recorded DRT and EEG simultaneously; Group 2 (n=1 amputee + n=10 non-amputees) recorded ECG and pupillometry. Both groups completed NASA-TLX. Measures could not all be recorded concurrently, motivating the split design.
Sample size: Group 1: n=11 (1 amputee, 10 non-amputee). Group 2: n=11 (1 amputee, 10 non-amputee). Total N=22.
Quality score: 5 / 5