Artificial intelligence (AI) integrated with bioinspired design enables the development of materials that adapt and dynamically respond to biological cues. In this study, a Drosera capensis-inspired thermo-responsive microneedle (MN) platform was developed, integrating motion, surface, and functional mimicry through AI-guided 4D printing. Shape memory polymers (SMPs) composed of tert-butyl acrylate (tBA) and 1,6-hexanediol diacrylate (HDDA) were designed to exhibit reversible shape recovery upon thermal stimulation. The complex shape-memory behavior was quantitatively modeled using multiple machine learning (ML) algorithms, including support vector regression (SVR), extreme gradient boosting (XGB), and Gaussian process regression (GPR). Among them, GPR demonstrated superior predictive accuracy (R2 > 0.99) and provided predictive means and 95% confidence intervals, highlighting its reliability in modeling nonlinear thermal recovery behavior and its potential for guiding process parameter optimization. The resulting MNs exhibited Drosera capensis-like coiling and grasping motions, enabling self-actuating wound closure. Furthermore, the MNs were functionalized with adhesive DNA (aDNA) and Zn nanolayers via sputtering-based plasma immersion ion implantation (S-PIII). The Zn nanolayers facilitated sustained DNA release and endowed the MNs with intrinsic antibacterial activity. In diabetic wound models, the AI-optimized biomimetic MN (BMMN) system significantly enhanced epithelial regeneration, collagen remodeling, and neovascularization, demonstrating adaptive and intelligent wound healing materials.
Lee et al. (2026) studied this question.