Analyzing Walking with Ankle Foot Orthoses Using Shank-mounted Wearable Movement Sensors
Cliona Blackwell · 2023 · ASSETS 2023: The 25th International ACM SIGACCESS Conference on Computers and Accessibility · doi:10.1145/3597638.3615654
Summary
This extended abstract investigates using Inertial Measurement Units (IMUs) mounted on the shank (lower leg) to provide quantitative gait analysis for clinicians tuning Ankle Foot Orthoses (AFOs). AFOs are orthotic devices that encompass the ankle joint and foot to improve gait in people with muscle deficiency, but the tuning process currently relies almost entirely on visual observation, with 87% of clinicians using visual inspection alone. This means clinicians do not capture a true picture of the patient's gait, particularly outside the clinic. The study collected data from one adult male volunteer who was a regular bilateral AFO wearer, using six NGIMU (x-io Technologies) sensors placed on the chest, waist, both thighs, and both shanks. Only the shank sensor data was analyzed. Four data collection sessions were performed: AFO with no heel wedges, AFO with 12mm heel wedges, AFO with 24mm heel wedges, and no AFO. The IMU quaternion data was converted to Shank Vertical Angle (SVA) using a rotation matrix, and a weighted k-nearest neighbors classifier (98% accuracy, 5-fold cross-validation) was used to automatically segment walking from non-walking periods. From the SVA signal, multiple gait metrics were extracted including angle metrics at heel strike, midstance, and toe off, as well as time-based metrics for step, stride, stance, and swing durations, plus a symmetry index. Formative interviews with two orthotists (6 months and 7 years experience) confirmed clinical interest in the sensor approach and validated the proposed metrics.
Key findings
The IMU sensors were sensitive to differences between AFO and non-AFO conditions across both angle- and time-based gait metrics. With AFOs (Session 1), the SVA range from toe off to heel strike was wider than without AFOs (Session 4), and the volunteer planted his heel further from his central line of body when wearing AFOs. Duration metrics were similar between conditions since the volunteer walked at approximately the same speed, though the AFO improved symmetry in stride and swing durations while reducing symmetry for stance and step durations. The paper proposes a novel metric called "negative momentum," defined as any backward movement of the shank during midstance, detected by a positive-to-negative gradient transition in the SVA signal. Negative momentum occurred more frequently with AFOs (1 in 8 steps) than without (1 in 9 steps), particularly during uphill walking where the volunteer needed to hyperextend his knee. This metric is clinically relevant because it captures inefficient gait that may indicate suboptimal AFO tuning. The automatic classification had limitations, misclassifying leg raises and transitional steps as walking, resulting in 163 steps detected automatically versus 142 manually for Session 1. The study demonstrates that two shank-mounted IMUs can be sufficient for meaningful gait analysis without a full gait laboratory.
Relevance
This research addresses a practical gap in assistive device fitting by proposing an affordable, portable alternative to expensive gait analysis laboratories. For accessibility practitioners and rehabilitation professionals, the approach could enable objective, data-driven AFO tuning in real-world environments rather than relying solely on subjective clinical observation in corridors. The novel negative momentum metric offers a potentially useful clinical indicator for identifying poorly tuned AFOs. However, the study has significant limitations: data from only one volunteer, analysis of only two of four sessions, and classification errors that affected metric calculations. The orthotist interviews revealed practical concerns about patient acceptance of wearing additional sensors alongside AFOs. Future work should validate the approach with multiple participants across different AFO types and conditions, address sensor placement consistency, and develop clinician-friendly data visualization tools. The work contributes to the broader goal of making rehabilitation technology assessment more accessible and evidence-based.
Tags: ankle foot orthosis · gait analysis · inertial measurement unit · wearable sensors · biomechanics · orthosis tuning · physical disability · mobility