Multiscale analysis of oxidative damage in hair fibers: From AFM nanomechanics to AI-based degradation modeling

Human hair is a hierarchical biomaterial whose mechanical performance depends on its cuticle integrity and keratin-based matrix. However, oxidative dyeing can compromise these nano-architectures. In this study, we employed scanning electron microscopy (SEM) and atomic force microscopy (AFM) with PinPoint nanomechanical mapping to investigate structural and mechanical changes induced by oxidative dyeing. SEM revealed progressive cuticle degradation—lifting, fraying, and intercellular voids—while AFM analysis showed reductions in Young's modulus and the emergence of localized softening (<3 GPa) with prolonged dye exposure. Quantitative mapping indicated increased surface roughness, density of soft regions, and mechanical heterogeneity as a function of exposure time. To simulate degradation dynamics, we used a conditional generative adversarial network (cGAN) trained on time-labeled morphological maps, which successfully interpolated intermediate states and reproduced experimental trends. Our integrated approach reveals the dose-dependent impact of dyeing on hair's protective structure and provides a multiscale framework—combining experimental nanomechanics and artificial intelligence modeling—for evaluating chemical damage in soft biomaterials.