An efficient detection method for camel milk quality based on dielectric spectroscopy technology.

Area of Science:

• Food Science and Technology
• Analytical Chemistry
• Machine Learning Applications

Background:

• Camel milk quality assessment faces challenges in detecting adulteration and nutritional content.
• Dielectric properties offer a potential non-destructive method for milk analysis.

Purpose of the Study:

• To analyze dielectric variations (dielectric constant ε' and dielectric loss factor ε") in camel milk components (fat, protein, lactose).
• To develop and compare machine learning models for detecting camel milk adulteration and quantifying nutritional components.

Main Methods:

• Dielectric spectroscopy was used to analyze camel milk samples across a 0.1 to 26.5 GHz frequency range.
• Detection models including Partial Least Squares (PLS), Convolutional Neural Network (CNN), Long Short-Term Memory (LSTM), CNN-LSTM, and an optimized CPO-CNN-LSTM-SEN were constructed.
• Model performance was evaluated for adulteration discrimination and simultaneous quantitative determination of fat, protein, and lactose.

Main Results:

• Dielectric properties (ε' and ε") showed strong negative correlations with fat, protein, and lactose content.
• CNN-LSTM achieved 100% accuracy in discriminating adulterated camel milk.
• The optimized CPO-CNN-LSTM-SEN model demonstrated high accuracy (R_P² up to 0.9703) for simultaneous quantitative determination of nutritional components.

Conclusions:

• Dielectric spectroscopy combined with advanced machine learning models offers a robust solution for camel milk quality control.
• The developed models can effectively detect adulteration and accurately quantify key nutritional components in camel milk.