Advances in Machine Learning for SERS Analysis

Surface-enhanced Raman scattering (SERS) has advanced significantly since the initial report in 1974. With the current substrates, characterization methods, and measurement systems, SERS has great potential as an analytical tool, and this potential is amplified with machine learning-based analysis of the measured spectra. This chapter will review the basics of machine learning relevant for SERS analysis, including underlying principles, the commonly used methods, and consideration of validation and figures of merit. For context, the chapter will review a range of specific examples where various ML techniques have been used for SERS analysis, finishing with an in-depth consideration of ML-enabled SERS detection of bacteria. Thoughtful ML application is an excellent way to maximize the utility of this sensitive and selective spectroscopic tool.