Data-driven characterization of walking after a spinal cord injury using inertial sensors.
Charlotte Werner, Meltem Gönel, Irina Lerch and 2 others
PMID 37120519WHAT IT FOUND
Ankle sensors during a six-minute walk identified four distinct gait patterns in incomplete spinal cord injury patients.
Adding these sensor metrics to clinical data improved the accuracy of predicting who would improve their walking capacity in future rehab sessions from 70% to 80%.
Key findings
01Clustering of sensor-derived gait parameters separated patients into four groups with distinct walking characteristics, such as compensatory movements or variability in double support.
02A prediction model using only clinical data (current walking distance and time since injury) achieved 70% accuracy in predicting whether a patient would improve their walking capacity.
03Including sensor-derived gait parameters in the prediction model improved the accuracy to 80%.
STILL TO COME
How it was doneWhat they foundWhat it means for PTs
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 is observational and does not test whether acting on the proposed cluster-specific recommendations improves patient outcomes. Many gait parameters are correlated with walking speed, making it difficult to separate quality deficits from speed deficits. The prediction model was trained on a small subset of 23 patients with longitudinal data (40 observations), limiting generalizability. Sensor-derived parameters are subject to estimation errors, although the authors state these were minimized by using a validated algorithm.
Declared interests
The study was funded by the Swiss Federal Institute of Technology Zurich. No other conflicts of interest are reported.
The easy way to misread this
Do not assume that the 80% prediction accuracy or the four gait clusters are ready for clinical use. The prediction model was trained on a very small sample of 23 patients with longitudinal data, and the clinical recommendations for each cluster were proposed by the authors but not tested in a trial.