Toward smart diagnostics via artificial intelligence-assisted surface-enhanced Raman spectroscopy

Molecular information contained in bodily fluids (ex. Blood, urine, saliva, or tears) can be minutely obtained through label-free surface-enhanced Raman spectroscopy (SERS). However, the resulting SERS spectra require complex analysis to transform such spectral information into accurate diagnostics. Herein, we review how scientists and technologists are employing SERS and artificial intelligence (AI) to carry out prediction, classification, spectral variation detection, and pattern recognition tasks to extract molecular information and generate diagnostic models or staging platforms based on disease-related molecular variations (reflected in the analyzed spectra). The employed SERS substrates are critically discussed and AI methods applied to assist SERS-based diagnostics are also elaborated. Particular applications such as the AI-assisted diagnosis of cancer, infectious diseases, and other illnesses (including stroke and Alzheimer's) are also covered. Besides, our perspective to push forward the frontiers of this exciting field toward smart diagnostics and their clinical translation is offered.