Bacterial Wastewater-Based Epidemiology Using Surface-Enhanced Raman Spectroscopy and Machine Learning.

Although wastewater-based epidemiology has been used extensively for the surveillance of viral diseases, it has not been used to a similar extent for bacterial diseases. This is in part owing to difficulties in distinguishing pathogenic from nonpathogenic bacteria using PCR methods. Here, we show that surface-enhanced Raman spectroscopy (SERS) can be a scalable, label-free method for the detection of bacteria in wastewater. We enhance Raman signal from bacteria in wastewater using plasmonic gold nanorods (AuNRs) that electrostatically bind to the bacterial surface and confirm this binding using cryoelectron microscopy. We spike four clinically relevant bacterial species and AuNRs into filtered wastewater, varying the AuNR concentration to maximize the signal. We then collect 540 spectra from each species at 10 cells/mL and train a machine learning model to identify them with more than 87% accuracy. We also demonstrate an environmentally realistic limit of detection of 10 cells/mL. These results are a key step toward a SERS platform for bacterial WBE.