Nutritional Characteristics of Foods With Addictive Potential: A Machine-Learning Approach

Area of Science:

• Nutritional Science
• Food Science
• Public Health

Background:

• Ultraprocessed foods (UPFs) dominate the US food supply.
• The perceived addictive potential of foods is a growing public health concern.
• Understanding the nutritional drivers of food addiction is crucial for dietary guidelines.

Purpose of the Study:

• To pinpoint nutritional characteristics linked to the perceived addictive nature of common US foods.
• To investigate the role of ultraprocessed foods (UPFs) in food addiction.
• To develop a framework for identifying foods with high addictive potential.

Main Methods:

• Surveyed US adults (n=1664) on the perceived addictiveness of 297 foods (74.4% UPFs).
• Utilized machine learning to analyze nutritional predictors of perceived addictiveness.
• Examined 15 standard Nutrition Facts label variables and 166 expanded nutrient characteristics.

Main Results:

• Nonlinear associations were found between nutrient content and perceived addictiveness.
• Foods rich in carbohydrates, glycemic load, energy density, and fat were rated as more addictive.
• UPFs frequently exhibited multiple addictive nutrient thresholds, unlike minimally processed foods.

Conclusions:

• A distinct nutritional signature correlates with perceived food addictiveness.
• Findings offer a data-driven approach to identifying foods promoting compulsive consumption.
• Results can inform policies for a healthier, less addictive food environment.