Machine learning-driven flavoromics: Decoding stage-specific volatile compound dynamics and sensory deterioration in...

Area of Science:

• Food Science
• Analytical Chemistry
• Computational Biology

Background:

• Infant formula (IF) is essential for infant nutrition.
• Storage causes sensory deterioration, affecting consumer acceptance.
• Understanding IF degradation mechanisms is crucial for quality control.

Purpose of the Study:

• Investigate stage-specific sensory deterioration in IF.
• Identify mechanisms driving flavor loss during storage.
• Develop predictive models for IF shelf-life optimization.

Main Methods:

• Flavoromics and fatty acid analysis.
• Quantitative descriptive analysis (QDA).
• Machine learning (Random Forest, XGBoost).

Main Results:

• Six attributes identified as key drivers of IF quality loss.
• Aldehydes and ketones identified as dominant off-flavor markers.
• Polyunsaturated fatty acid degradation correlated with off-flavor generation.

Conclusions:

• Lipid oxidation is the primary mechanism for IF flavor deterioration.
• Predictive models accurately link fatty acid oxidation to sensory decline.
• Developed a data-driven tool for optimizing IF shelf-life.