S2 - A Multiplexed SERS-biosensor Assay for the Discrimination of Diseases with Overlapping Symptoms

Sunday, October 5, 2025

7:15 PM - 9:15 PM EDT

Location: Ballroom C

Presenting Author(s)

SF

Samson A. Fashanu

Graduate Student
University of Cincinnati
Cincinnati, Ohio, United States

Non-Presenting Author(s)

Steven M. Quarin (he/him/his)

Graduate Research Assistant
University of Cincinnati
Cincinnati, Ohio, United States

Project Lead(s)

PS

Pietro Strobbia, PhD

Assistant Professor
University of Cincinnati
Cincinnati, Ohio, United States

Submitter(s)

SF

Samson A. Fashanu

Graduate Student
University of Cincinnati
Cincinnati, Ohio, United States

In fifteen words or less, explain the significance of this contribution (Novel Aspect).: Multiplexed detection of biotargets with reagentless SERS-biosensors

Abstract Text: The global pandemic emphasized the need to develop sensitive, affordable and easy-to-use methods for detecting biomarkers; however, the current standard testing methods can be limited in the number of biomarkers simultaneously detected in a single diagnostic test. To expand these capabilities, we have designed a multiplexed sensor to detect different respiratory diseases with overlapping symptoms such as Influenza A, Respiratory Syncytial Virus, SARS-CoV-2, SARS-CoV-2 omicron variant and human RNase serving as a control. These reagentless SERS sensors are ideal for point-of-care analysis because they work in a one-pot assay format, with no sample preparation needed, while having improved multiplexing capabilities. Due to the presence of multiple reporters for different biomarkers, overlapping spectra become complex to resolve, making machine learning key for spectral unmixing for quantification of targets. To this end, we tested various options, including PCA, PCR and PLS. Using PLSR each individual sensor from the multiplex was quantified independently and we were able to successfully quantify components of unknown mixtures. The results obtained demonstrate the capabilities of SERS sensing in conjunction with machine learning. Further research is ongoing in order to improve the robustness of the model and improve its practicality for point of care diagnostics.