Classification of left and right-hand motor imagery in acute stroke patients using EEG microstate.
Shiyang Lv, Xiangying Ran, Mengsheng Xia and 6 others
PMID 40533772WHAT IT FOUND
EEG brain activity patterns during imagined left or right hand movements differed significantly in 14 acute stroke patients.
A machine learning model correctly classified the side of imagined movement in 75% of trials. These distinct neural signatures could help guide future brain-computer interface rehabilitation tools.
Key findings
01Microstate A parameters (duration, occurrence, coverage) were significantly higher during left-hand motor imagery, while microstate C parameters were significantly higher during right-hand motor imagery.
02A K-Nearest Neighbors (KNN) machine learning model using these microstate features achieved a classification accuracy of 75.00% for distinguishing left versus right-hand motor imagery.
03The occurrence of microstate C was identified as the most important feature for distinguishing between left and right-hand motor imagery tasks.
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 included only 14 patients from a single dataset, limiting generalizability. The sample was heavily skewed toward males (12 of 14), which may influence results. Heterogeneity in stroke lesion location and severity among participants could affect brain activity patterns. The classification accuracy of 75% is modest and may not be sufficient for reliable real-time clinical use.
The easy way to misread this
Do not interpret the 75% classification accuracy as evidence that this technology is ready for clinical rehabilitation. The study used a very small, single-center sample with high variability in stroke location and severity, and the accuracy was not significantly different from chance-level performance for some classifiers.