Zinc oxide/polydimethylsiloxane‐coated carbon fiber yarns for machine‐learning‐assisted gait classification

Abstract Gait pattern analysis is particularly valuable for the early diagnosis of neurological disorders such as dementia and Parkinson's disease; however, comprehensive multidimensional evaluation methods remain limited. In this context, intelligent sensing platforms are gaining prominence for continuous physiological monitoring, providing real‐time data for healthcare and rehabilitation. Within this scope, this work presents a method to produce zinc oxide/polydimethylsiloxane (ZnO/PDMS) coated carbon fiber yarns, which can serve as building blocks for smart textile sensing platforms. Zinc oxide wires were synthesized via a cost‐ and time‐efficient microwave‐assisted hydrothermal process, while in situ Joule heating enabled the formation of uniform coatings around the conductive yarns. Among the tested ZnO concentrations, 1–20 wt.%, fibers with 10 wt.% ZnO yielded the highest output voltage of (34 ± 1) V under a compression force of 30 N. The pressure‐sensing performance of these fibers was tested under compressive forces between 30 and 60 N, yielding sensitivity values of (0.62 ± 0.08) V N −1 and (0.049 ± 0.004) μA N −1 . The optimized ZnO/PDMS‐coated carbon fibers were integrated into a smart sock and tested for gait‐pattern recognition and gait abnormality detection using machine learning algorithms. Support vector machine achieved the best performance for gait pattern detection, with an accuracy of 0.928 and a Matthews correlation coefficient (MCC) equal to 0.893 when distinguishing walking, marching and running scenarios. For gait abnormality detection, the random forest classifier yielded the best results, reaching an accuracy of 0.894, an area under the curve of 0.936 and an MCC value equal to 0.761. The proposed approach combines scalable fabrication with machine learning, providing a cost‐effective and versatile platform for self‐powered, textile‐based wearable devices with potential applications in human activity recognition, gait analysis, medical monitoring and rehabilitation.