Detecting compensatory movements of stroke survivors using pressure distribution data and machine learning algorithms.
Siqi Cai, Guofeng Li, Xiaoya Zhang and 4 others
PMID 31684970WHAT IT FOUND
In eight stroke survivors, a pressure-distribution mattress and machine learning classified trunk lean-forward, trunk rotation and shoulder elevation during seated reaching, and the patterns matched trunk muscle activity.
Key findings
01The k-NN and SVM classifiers separated compensatory from normal movement with all F1-scores above 0.95 and an average F1-score of 0.993.
02The classifiers also performed multiclass recognition of normal movement and three compensation patterns with an average F1-score of 0.981.
03Trunk muscle activity differed significantly between healthy and affected sides (P = 0.027), and the sEMG patterns matched the pressure-based classification.
STILL TO COME
How it was doneWhat they foundWhat it means for PTsWhat it means for OTs
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
Only eight stroke survivors were included, and they had fair to good cognition and enough motor ability to sit and reach, so the results may not apply to more impaired stroke survivors. Classifier performance was evaluated by cross-validation within the same group, not by testing on a separate new sample. The system detected whether compensation was present and which pattern dominated, but it did not measure how much compensation occurred. The paper did not test whether pressure-based feedback or trunk restraint improves upper-limb function. Trunk muscle activity varied greatly between patients, so one classification model may not fit all movement styles.
Declared interests
Funding was provided by the National Natural Science Foundation of China, Key Technologies Research and Development Program, Natural Science Foundation of Guangdong Province, and Guangzhou Research Foundation. No additional author conflicts are stated.
The easy way to misread this
Do not read the high classifier scores as evidence that this system improves stroke rehabilitation. It only tested whether pressure data and machine learning could label movements in a small group, not whether feedback or trunk restraint changed function.