Dietary Pattern Extraction Using Natural Language Processing Techniques.

Area of Science:

• Nutrition Science
• Public Health
• Computational Methods

Background:

• Dietary patterns significantly impact population health.
• Understanding shifts in dietary habits is crucial for public health interventions.
• Previous analyses relied on traditional methods for dietary pattern extraction.

Purpose of the Study:

• To analyze changes in Korean adult dietary patterns over a decade.
• To compare dietary patterns from 2007-2009 with those from 2016-2018.
• To explore novel computational methods for dietary pattern analysis.

Main Methods:

• Utilized 24-hour dietary recall data from the Korea National Health and Nutrition Examination Survey (KNHANES).
• Applied machine learning and natural language processing (NLP) techniques for dietary pattern extraction.
• Identified three distinct dietary patterns for each survey period.

Main Results:

• Observed a significant increase in Western dietary patterns among Korean adults between 2007-2009 and 2016-2018.
• Identified dietary patterns comparable to traditional and Western styles in both periods.
• Demonstrated the efficacy of NLP in extracting detailed dietary information.

Conclusions:

• Korean adult diets have shifted towards Western patterns over the past decade.
• Natural language processing offers a powerful, novel approach for dietary pattern analysis.
• Findings provide valuable insights for targeted public health nutrition strategies in Korea.