Predicting Unreported Micronutrients From Food Labels: Machine Learning Approach.

Area of Science:

• Computational nutrition and public health informatics.
• Application of machine learning in dietary assessment.

Background:

• Micronutrient deficiencies affect over 2 billion people globally, posing a significant public health challenge.
• Existing food labels offer limited information on vitamins and minerals due to regulatory and physical constraints.
• This gap in information hinders consumers' ability to make informed dietary choices.

Purpose of the Study:

• To investigate the use of machine learning algorithms for predicting unreported micronutrients (vitamins and minerals) from standard food label data.
• To assess the accuracy of predictive models in classifying micronutrient levels (low, medium, high).
• To explore the potential integration of these models into mobile applications for consumer benefit.

Main Methods:

• Utilized the Food and Nutrient Database for Dietary Studies (FNDDS) dataset, comprising 5624 food items.
• Trained diverse machine learning classification and regression algorithms to predict micronutrient content.
• Employed hyperparameter tuning and repeated cross-validation to ensure model robustness and prevent overfitting.

Main Results:

• Regression models showed variable accuracy in predicting exact micronutrient quantities (R² from 0.28 to 0.92).
• Classification models achieved high accuracy (exceeding 0.80) in categorizing micronutrient levels (low, medium, high).
• Top classification accuracies were observed for vitamin B12 (0.94) and phosphorus (0.94), with vitamin E (0.81) and selenium (0.83) showing lower, yet still strong, performance.

Conclusions:

• Machine learning effectively predicts micronutrient categories from existing food label information, demonstrating feasibility.
• This approach can significantly enhance consumer awareness of food's micronutrient profiles.
• Integration into mobile apps offers a scalable solution for personalized dietary guidance and improved public health outcomes.