AI-guided 3D bioprinting of multi-drug-loaded chitosan methacrylate hydrogels for personalized diabetic ulcer repair.

Chronic diabetic foot ulcers (DFUs) are plagued by irregular geometries, persistent inflammation, and poor angiogenesis, demanding personalized therapeutic strategies. Herein, we developed a closed-loop platform integrating AI multi-scale fusion segmentation, 3D bioprinting drug-stratified hydrogels and diabetic-specific wound modeling to address DFU management. The AI workflow combined YOLOv8 for wound segmentation and Depth Anything V2 for depth estimation, optimized via multi-scale prediction fusion to enhance robustness. A refined dataset (210 lean pork wound images +180 clinical DFU images) replaced non-physiological fatty tissue, with external validation on 50 clinical wound samples (Dice coefficient: 0.93 ± 0.02). Chitosan methacrylate (CsMeA) hydrogels loaded with tailored drug combinations (VEGF/PDGF for angiogenesis, AgSD/honey for infection control, LIDHCl/LVX for pain/inflammation) were 3D-bioprinted with layer-specific drug deposition. In vitro, the hydrogels exhibited sustained release (up to 14 days), >85% cell viability, and synergistic antibacterial activity (inhibition zone increase: 61%and 81%for LVX against S. aureus and P. aeruginosa). In vivo studies on diabetic pigs showed VEGF/honey hydrogels achieved 93.62% ± 1.83 wound closure at Day 21, significantly higher than non-personalized controls (72.3 ± 3.5%), with enhanced angiogenesis and reduced inflammation. This platform advances personalized DFU care by bridging AI-driven diagnostics and targeted therapeutics, with improved translational potential.