Intelligent wearable system design for personalized knee motion and swelling monitoring in osteoarthritis care

Abstract

Daily knee monitoring is critical for osteoarthritis management, aiding in both prevention and rehabilitation. Current wearable solutions for daily use typically capture knee-bending angles as a single feature but lack evidence for comprehensive knee-state recognition. Here we introduce SyncKnee, a knee-monitoring system that tracks both joint angles and swelling patterns, providing detailed knee-state monitoring for daily use. SyncKnee consists of three components: a stretch sensor pad, a multi-modal machine-learning model, and personalized information support. The sensor, made from poly(SBS) fiber and eutectic gallium-indium alloy, tracks skin deformation from bending and swelling. Robotic-arm-driven tests confirm sensor accuracy in responding to bending and swelling. In the user study with 15 participants performing five distinct knee maneuvers, our system with a random forest model achieves 98.48% accuracy in recognizing knee behaviors. SyncKnee offers a comprehensive approach to knee monitoring with promising applications for daily osteoarthritis care.