Machine learning supported single-stranded DNA sensor array for multiple foodborne pathogenic and spoilage bacteria...

Area of Science:

• Food Science and Technology
• Biosensing and Diagnostics
• Microbiology

Background:

• Ensuring food safety requires rapid and accurate detection of pathogenic bacteria.
• Current methods like plate counting and ELISA can be time-consuming or labor-intensive.
• Contaminated milk samples pose a significant risk in the food supply chain.

Purpose of the Study:

• To develop a non-specific optical sensor array for identifying multiple pathogenic bacteria in food.
• To create a rapid, accurate, and cost-effective alternative to existing bacterial detection methods.
• To establish a unique fluorescence fingerprinting technique for bacterial species identification.

Main Methods:

• Utilized fluorescence-labeled single-stranded DNA quenched by 2D nanoparticles.
• Employed recovery of fluorescence by foreign biomolecules to generate bacterial fingerprints.
• Applied machine learning models, including artificial neural networks, for classification.
• Tested sensor performance with contaminated milk samples spiked with eight bacterial species.

Main Results:

• The sensor array successfully identified eight different bacterial species within hours.
• An artificial neural network achieved 93.8% accuracy with a 30-minute incubation period.
• Extending incubation to 120 minutes improved the multiplayer perceptron accuracy to 98.4%.

Conclusions:

• The developed optical sensor array offers a novel, low-cost, and high-accuracy approach for bacterial identification.
• This method provides a viable alternative to traditional bacterial detection techniques.
• The fluorescence fingerprinting approach enables rapid and reliable detection of multiple bacteria in food samples.