Automated assessment of upper limb spasticity in stroke patients with fusion of multichannel surface electromyography features.
Xin Li, Feiyu Nong, Xu Zhang and 4 others
PMID 40874349WHAT IT FOUND
A machine learning model using muscle sensors classified spasticity severity in 40 stroke patients with 78.7% overall accuracy.
It distinguished mild from severe cases well but failed to separate moderate grades, meaning it cannot yet replace clinical assessment.
Key findings
01The multichannel model achieved 78.7% average accuracy, which was higher than single-channel or single-feature models.
02The model accurately identified the most severe spasticity (MAS 3) with 83% recall, but performed poorly on moderate grades (MAS 1+ and MAS 2) with F1 scores below 0.7.
03Time-domain features (RMS, iEMG, EA) correlated positively with spasticity grades, while frequency-domain features were negatively correlated and excluded.
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
The study included only male patients, so results may not apply to female patients. The sample size was very small (40 patients analysed), limiting generalisability. The model failed to accurately classify moderate spasticity grades (MAS 1+ and 2), which are critical for many clinical decisions. The study did not compare the model against healthy controls or lower limb data. The reference standard (MAS) is subjective and known to have poor inter-rater reliability, which may have introduced noise into the training data.
Declared interests
The authors declared no conflicts of interest. The study was funded by Guangxi Natural Science Foundation and Medical and Health Key Discipline Construction projects.
The easy way to misread this
Do not interpret the 78.7% overall accuracy as evidence that this tool is ready for clinical use. The model performs poorly on moderate spasticity grades (MAS 1+ and 2), which are the most common and clinically challenging levels to assess. Using it could lead to incorrect grading of these patients.