Title:

Smart Insole System for Detecting Lower-Limb Malalignment

Knee malalignment conditions such as genu varum (bow legs) and genu valgum (knock knees) are commonly assessed through clinical observation or costly imaging. This work presents SmartInsole, a proof of concept system for automated knee alignment classification using plantar pressure insole data. A 96 sensor pressure insole captures ground reaction force distributions during walking, which are processed into bilateral foot heatmap images using Gaussian splatting and stance phase averaging across full gait cycles. A ResNet18 convolutional neural network, pretrained on ImageNet and fine-tuned on a synthetically generated dataset of 558 heatmap images, classifies pressure patterns into three categories: normal alignment, bow legs, and knock knees. The trained model is converted to Apple CoreML format and deployed within a native iOS application, which displays the pressure heatmap, outputs a classification result, and provides condition specific rehabilitative exercise guidance. The system demonstrates a viable pipeline from raw sensor data to actionable clinical feedback on a mobile platform, with future work targeting real patient data collection and live Bluetooth sensor integration.

Name

Comment*

Comments are viewable only by submitter

Captcha*