Rapid analysis of meat floss origin using a supervised machine learning-based electronic nose towards food...

Area of Science:

• Food Science
• Analytical Chemistry
• Sensor Technology

Background:

• Consumer demand for authentic meat floss is high due to allergy and religious concerns.
• Accurate origin authentication is crucial to prevent food fraud and ensure consumer safety.
• Existing methods for meat floss authentication can be time-consuming and costly.

Purpose of the Study:

• To develop and evaluate a compact portable electronic nose (e-nose) for classifying different meat floss products.
• To assess the performance of various supervised machine learning algorithms for meat floss discrimination.
• To validate the e-nose findings using established analytical techniques like FTIR and GC-MS.

Main Methods:

• Development of a portable e-nose system with a gas sensor array.
• Application of supervised machine learning algorithms (LDA, QDA, k-NN, RF) with window time slicing.
• Classification of beef, chicken, and pork floss samples.
• Correlation of e-nose data with Fourier-transform infrared (FTIR) spectroscopy and gas chromatography-mass spectrometry (GC-MS) results.

Main Results:

• The linear discriminant analysis (LDA) model achieved >99% accuracy in discriminating between beef, chicken, and pork floss.
• Beef and chicken floss samples showed similar compound profiles (hydrocarbons, alcohol).
• Pork floss samples were characterized by dominant aldehyde compounds, such as dodecanal and 9-octadecanal.

Conclusions:

• The developed portable e-nose system demonstrates high accuracy and potential for reliable meat floss authentication.
• This technology offers a rapid and effective tool for detecting food fraud and ensuring product authenticity.
• The findings support the widespread application of e-nose technology in food authenticity testing.