Design, development, and evaluation of a local sensor-based gait phase recognition system using a logistic model decision tree for orthosis-control.

All participants were able-bodied, so the study did not show how the system performs in people with knee-extensor weakness, the group stance-control knee orthoses are meant to help. The data came from a motion-capture laboratory system, not from the actual sensors that would be used in an orthosis. The authors noted that real sensor data may be noisier and lower quality, so the classification results may not transfer directly. The paper reports classification performance, not patient walking outcomes, falls, knee stability, or actual orthosis control. Misclassifications occurred at phase transitions, especially around push-off and loading response in validation without correction. Very slow walking was difficult because signal variation was small.

The study was funded by the Natural Sciences and Engineering Research Council of Canada. No other conflicts of interest are reported in the supplied text.

Do not conclude that this system is ready for patients or improves mobility. It was tested only on able-bodied motion-capture data, not in actual orthoses or people with knee weakness.