Presentation
Classifying Restricted Knee Flexion and Affected Sides Using Plantar Pressure Data During Walking: A Preliminary Study
DescriptionDiminished knee flexion during walking, commonly observed in neuromuscular and knee-related diseases, compromises gait stability and increases the risk of falls. This study developed an accessible, non-invasive algorithm using pressure platform data to predict restricted knee flexion and affected sides. Sixty healthy young adults performed walking trials on a 10m path with a 1.5m pressure platform, under both normal and restricted conditions. Their left and right knees were constrained to 20 and 40 degrees using a knee brace in a randomized order. We employed machine learning techniques, including feature selection, stratified 10-fold cross-validation, and hyperparameter optimization, on datasets split into 80% training and 20% testing sets. The optimized Support Vector Machine distinguished between normal and restricted conditions and identified the affected side with 95% accuracy. This preliminary study provides valuable insights and could serve as a practical tool for enhancing mobility assessments in patients with neuromuscular and knee-related conditions.
Event Type
Lecture
TimeTuesday, September 10th1:30pm - 1:50pm MST
LocationGrand Ballroom
Health Care