Skip to content

Machine-Learning-Assisted Raman Spectroscopy for Rapid Identification of Dominant Bacterial Species in Five Community State Types of Human Vaginal Fluids

Oct 2026 · ACS Applied Nano Materials · 0 citations · 69 references

Abstract

Bacterial vaginosis (BV), the most prevalent vaginal dysbiosis in women of reproductive age, is associated with a notably high rate of recurrence following treatment. Persistent BV increases the risk of sexually transmitted infections and negatively impacts female reproductive health. Currently, bacterial identification in vaginal samples primarily relies on methods such as conventional culture-based methods, nucleic acid amplification test, matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS), and high-throughput sequencing technologies, etc. However, these approaches are limited by long turnaround times and/or high costs, requiring the development of detection techniques. In this study, the non-invasive surface-enhanced Raman spectroscopy (SERS) technique was applied to characterize the spectral features of dominant bacteria in vaginal community state types (CSTs), including Lactobacillus crispatus (CST I), Lactobacillus gasseri (CST II), Lactobacillus iners (CST III), Lactobacillus jensenii (CST V), Gardnerella vaginalis (CST IV), and Fannyhessea vaginae (CST IV). Due to the data complexity and high dimensionality of the SERS spectra, regular statistical methods are insufficient for spectral analysis. Therefore, machine learning algorithms were developed to construct classification models for rapid bacterial identification. The linear discriminant analysis (LDA) model demonstrated the best performance, achieving a classification accuracy of 94.61% and an area under the receiver operating characteristic curve (AUC) of 0.9977. Furthermore, validation using independently acquired SERS spectra yielded a prediction accuracy of 80.49%. These findings demonstrate that the SERS technique combined with machine-learning models provides a promising strategy for rapid discrimination of dominant bacterial species associated with vaginal CSTs, with potential applications in future studies of vaginal microbial communities following further validation in complex and clinical samples.

View source

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.