Off‐Axis Strain‐Invariant, Soft Pressure Sensor for Adaptive Transfer Learning‐Aided Patellofemoral Motion Monitoring

Abstract Stretchable pressure sensors have exceptional adaptability for conformal‐attachment monitoring on curved biological surfaces, enabling mechanosensing applications such as patellofemoral motion monitoring. However, conventional soft pressure sensors face challenges from off‐axis deformations (eg, bending or stretching), compromising signal fidelity and limited adaptability to new subjects in posture prediction. Here, a stretchable iontronic pressure sensor with off‐axis strain‐insensitivity is presented. The sensor leverages the electric double‐layer capacitive sensing mechanisms to achieve superior sensitivity (18.174 kPa −1 ) while maintaining minimal inter‐unit crosstalk (<1% signal interference) in the array. Serving as the mechanical‐strain buffers, the microcavity spacer and ultra‐stretchable electrode mitigate passive strain concentration to make the sensor mechanically resilient against both bending (up to 0.67 cm −1 curvature) and tensile (≤40% strain) deformations, respectively. The stretchable sensor array continuously monitors dynamic pressure over curved patellofemoral skin during knee motions. Via an effective transfer learning algorithm, continuous and nonlinear 16D pressure data achieves >95% classification accuracy corresponding to the subjects’ patellofemoral motion, which also adapts data from a new subject to predict the unknown posture. The off‐axis‐strain invariant and stretchable pressure sensor with transfer learning‐based adaptive capability holds potential for knee injury evaluation, dynamic rehabilitation, and real‐time biomechanical analysis.