Application of Inertial Measurement Units and Machine Learning Classification in Cerebral Palsy: Randomized Controlled Trial.
Siavash Khaksar, Huizhu Pan, Bita Borazjani and 7 others
PMID 34668870WHAT IT FOUND
Custom wrist sensors plus machine learning separated movement samples from children with CP from typical movement samples on a stop sign task, correctly classifying 85.75% of older samples and 88.13% of younger samples.
This is a measurement method, not proof that therapy works.
Key findings
01For the older group, the best classifier reached 85.75% accuracy, compared with 57.02% for the simplest baseline.
02For the younger group, a one-feature classifier reached 88.13% accuracy, compared with 53.44% for the simplest baseline.
03The sensors captured active wrist range of motion during functional tasks such as a stop sign motion.
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
This is a sensor and machine learning analysis, not a clinical trial outcome about whether orthoses help children with CP. The accuracy was measured on existing trial data using cross-validation, so it does not show that the method works in routine clinic use. The paper gives inconsistent statements about the best accuracy for the younger group. The sensor setup had practical problems, including drift, small hands, sensors falling off, therapist contact during goniometer use, corrupted data, and changes to the accelerometer scale. The classification was based on a stop sign task, so it says little about everyday hand movement.
Declared interests
The authors declared no conflicts of interest.
The easy way to misread this
Do not read the accuracy numbers as evidence that the sensors diagnose CP in clinic or that wrist and hand orthoses work. They come from classifying stop sign task samples in existing trial data.