Authentication of Edible Oil by Real-Time One Class Classification Modeling

Area of Science:

• Chemometrics
• Food Science
• Analytical Chemistry

Background:

• Adulteration detection is crucial for food safety and authenticity.
• One-class classification (OCC) models require representative samples, which are difficult to obtain globally.
• Evaluating sample representativeness for OCC models is challenging.

Purpose of the Study:

• To develop a novel authentication method for edible oils that bypasses the need for a pre-built prediction model.
• To identify adulterated oils in the market using real-time OCC modeling and model population analysis.
• To provide a new approach for detecting adulteration in high-value food products.

Main Methods:

• A new authentication method based on real-time one-class classification (OCC) modeling and model population analysis was developed.
• Numerous OCC models were generated using Monte Carlo sampling on subsets of inspected samples.
• The method identifies adulterated samples by analyzing the absolute centered residual (ACR) of models built with authentic versus mixed samples.

Main Results:

• The study successfully identified 6 out of 40 avocado oil samples as adulterated.
• The identified adulterants included soybean oil, corn oil, and rapeseed oil.
• Validation by chemical markers confirmed the effectiveness of the proposed method in detecting adulteration.

Conclusions:

• The proposed method offers a novel and effective approach for edible oil authentication without prior model building.
• This technique demonstrates significant potential for detecting adulteration in various high-value food products.
• The study validates the underlying philosophy that the ACR differs significantly between models of pure and adulterated samples.