OtherJournal of neuroengineering and rehabilitation2024

Multi-scale attention patching encoder network: a deployable model for continuous estimation of hand kinematics from surface electromyographic signals.

Chuang Lin, Qiong Xiao, Penghui Zhao

PMID 39736665

WHAT IT FOUND

A new AI model predicts hand movements from muscle signals with high accuracy and runs fast enough for wearable devices.

It outperformed existing methods in lab tests using data from 40 healthy people, but it has not been tested in patients with disabilities.

Key findings

01The new model achieved higher correlation and lower error rates than five existing deep learning methods when predicting 40 hand movements.

02The model ran on a Raspberry Pi with a latency of 97.93 ms, meeting the under-200 ms requirement for practical wearable use, whereas LSTM and sBERT exceeded this limit.

03The study used only healthy participants, so the model's performance for people with motor impairments remains unverified.

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.

Already have one?

What it does not show

The study was conducted entirely on healthy participants. There is no evidence that the model works for individuals with neuromotor impairments, such as those who might use prosthetics or rehabilitation devices. Validation relied on a single public dataset (Ninapro DB2) and simulated noise, rather than real-world clinical environments or diverse patient populations. The model was tested in a controlled laboratory setting; factors like electrode displacement, sweat, or long-term signal drift in daily life were not comprehensively evaluated.

Declared interests

Funding was provided by the Leading Talent Project of Dalian Maritime University. No commercial conflicts of interest were declared.

The easy way to misread this

Do not assume this technology is ready for clinical use with patients. The high accuracy reported here was achieved in healthy volunteers; its performance in individuals with muscle weakness, spasticity, or altered biomechanics has not been tested.

Read it on PubMed →


The study

Participants
40 healthy subjects
Certainty of evidence
Low

Browse

    Cite

    Chuang Lin, Qiong Xiao, Penghui Zhao Multi-scale attention patching encoder network: a deployable model for continuous estimation of hand kinematics from surface electromyographic signals. Journal of neuroengineering and rehabilitation. 2024.

    Read the original