Data-driven design of disordered structures for direction- and geometry-independent stretchable electrodes

Stretchable electrodes with strain-insensitive performance across arbitrary directions and geometries are essential for next-generation customizable wearable electronics. However, conventional designs based on ordered structures such as serpentines, kirigami, and island-bridge generally exhibit directional limitations and geometry constraints, restricting their applicability in complex scenarios. Here, we report a universal machine-learning-guided design strategy that generates bioinspired disordered structures for direction- and geometry-independent, strain-insensitive stretchable electrodes. By integrating a neural network model with an evolutionary algorithm, an optimization framework is established to discover disordered structures with enhanced stretchability and minimized resistance variation under unidirectional, bidirectional, and omnidirectional deformation. Remarkably, this approach also proves effective for irregular geometries, exemplified by hand-shaped electrodes. The optimized disordered structures fabricated via 3D printing achieve a 20–50% reduction in relative resistance compared with the ordered counterparts. Leveraging these electrodes, we further develop a novel stretchable wireless electroencephalography (EEG) cap that replaces conventional rigid, wired designs, accommodates diverse head sizes, suppresses motion artifacts, and achieves a high recognition accuracy in brain-computer interaction tasks. The successful demonstration highlights a robust and universal design paradigm for geometrically adaptive wearable electronics, advancing data-driven innovation in stretchable structures.