Abstract

Gait rehabilitation immediately after a patient has suffered a stroke is thought to be a very promising method to correct hemiplegic gait patterns. Delayed therapy often leads to secondary impairments such as disuse atrophy and general deconditioning that compounds the patient's stroke related ambulatory problems. A smart knee brace (SKB) has been developed as a training device to facilitate physical therapists in early gait training.

In this thesis, the hardware, electronics, and software design that comprise the brace are examined. The SKB features a fixed extension lock and a variable angle flexion lock. The extension lock is used in the beginning of stance phase to induce flexion and force the patient to utilize the quadriceps. The flexion lock is used to control excessive motion in the stance phase and prevent the patient from collapsing. To control the locking mechanism, the SKB control algorithm requires information based on sensors attached to the brace. The brace is fully controllable by a physical therapist through interface software programmed in MATLAB xPC target.

A real-time gait event detection algorithm was developed based solely on kinematic sensors on the brace as an alternative to footswitches. The current algorithm has advantages over footswitches that tend to be inaccurate, failure prone, and inconvenient for the standalone brace. The algorithm was applied to a healthy subject and to two stroke patients and was shown to accurately detect initial contact and toe off gait events.