Prediction of robotic neurorehabilitation functional ambulatory outcome in patients with neurological disorders.
Chao-Yang Kuo, Chia-Wei Liu, Chien-Hung Lai and 3 others
PMID 34922571WHAT IT FOUND
Early Lokomat settings predict gait recovery.
Patients who reduced body weight support below 34% and guidance force below 70% by session 6 were more likely to improve walking independence. This model aids planning but requires larger validation before clinical use.
Key findings
01Reducing body weight support to less than 34% by the 4th session and guidance force to less than 70% by the 6th session was associated with a higher likelihood of improving functional ambulation.
02A machine learning model using data from all training sessions accurately predicted which patients would improve their walking independence, with the Random Forest algorithm showing the highest performance.
03Patients who started training within three months of onset and those with a single affected extremity were more likely to show improvement in walking function.
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 used a small sample size (91 patients) from a single hospital, which limits generalizability. The dataset was artificially balanced by oversampling the non-improvement group, which may affect the real-world applicability of the model's sensitivity and specificity. The study is retrospective, relying on existing data without controlling for other concurrent therapies or patient adherence. The model was not externally validated on a new dataset, so its accuracy in other clinics is unknown. Only three Lokomat parameters were analyzed; other settings like range of motion or asymmetry were not included. The high cost of treatment in Taiwan may have selected for patients who could afford it, potentially skewing the population.
Declared interests
Funded by the Ministry of Science and Technology and the Ministry of Education in Taiwan. No specific commercial conflicts of interest were declared.
The easy way to misread this
Do not use this model to deny patients robotic therapy. The authors explicitly state the classifier is not meant to stop training but to aid in planning. The high accuracy reported is from internal cross-validation on a small, oversampled dataset, so it may not perform as well in a new, unbalanced population.