Detection of Helicobacter pylori Infection in Human Gastric Fluid Through Surface-Enhanced Raman Spectroscopy Coupled...

Area of Science:

• Biomedical Engineering
• Medical Diagnostics
• Machine Learning

Background:

• Current Helicobacter pylori diagnostic methods like urea breath tests and antibody assays have limitations including low accuracy and invasiveness.
• There is a significant need for simple, rapid, and noninvasive diagnostic tools for H. pylori infection.

Purpose of the Study:

• To develop and evaluate a novel, noninvasive method for detecting H. pylori infection using machine learning analysis of gastric fluid surface-enhanced Raman scattering (SERS) spectra.

Main Methods:

• Gastric fluid samples were noninvasively collected from 100 participants using the string test.
• 12,000 SERS spectra were generated and analyzed using machine learning algorithms, with the Light Gradient Boosting Machine (LightGBM) selected for its performance.
• The LightGBM model was validated on a separate set of 2,000 SERS spectra from 100 participants with unknown infection status.

Main Results:

• The LightGBM algorithm achieved a high prediction accuracy of 99.54% and time efficiency of 2.61 seconds in initial model building.
• Blind testing of the LightGBM model on unknown samples yielded a prediction accuracy of 82.15% when compared to qPCR results.

Conclusions:

• Machine learning analysis of SERS spectra from noninvasively collected gastric fluid presents a promising, simple, and rapid approach for H. pylori diagnosis.
• This novel technique has the potential to complement existing diagnostic methods for H. pylori infection.