Chenggang Zhang, Zeming Zhao, Meng Wang, Zhaotong Wang, Yuanlin Wang, Miao Yu, Ye Sun
Vibrio infections progress rapidly with high mortality, yet current diagnostics fail to meet the critical need for rapid, accurate, and field-deployable detection. To bridge this gap, we present a bioreceptor-free, AI-powered surface-enhanced Raman spectroscopy (SERS) platform for rapid Vibrio typing. Our approach leverages Na 2 HPO 4 -mediated synergistic anion/cation interactions to direct targeted accumulation of gold nanoparticles onto bacterial surfaces within 5 min, overcoming the uncontrollable aggregation and poor reproducibility of conventional methods. By integrating a gradient-weighted class activation mapping-assisted Vision Transformer model for intelligent spectral analysis with a custom-built portable Raman instrument for automated measurements, a remarkable typing accuracy of 97.62% is achieved across seven bacterial species (five Vibrio species and two interferents). The entire sample-to-result process takes only 10 min. This work establishes a new paradigm for SERS-based bacterial detection/typing and provides a promising solution for point-of-care diagnostics in public health, food safety, and environmental monitoring.