An Explainable Prediction for Dietary-Related Diseases via Language Models DOI Open Access
Insu Choi, Jihye Kim, Woo Chang Kim

et al.

Nutrients, Journal Year: 2024, Volume and Issue: 16(5), P. 686 - 686

Published: Feb. 28, 2024

Our study harnesses the power of natural language processing (NLP) to explore relationship between dietary patterns and metabolic health outcomes among Korean adults using data from Seventh Korea National Health Nutrition Examination Survey (KNHANES VII). Using Latent Dirichlet Allocation (LDA) analysis, we identified three distinct patterns: “Traditional Staple”, “Communal Festive”, “Westernized Convenience-Oriented”. These reflect diversity preferences in reveal cultural social dimensions influencing eating habits their potential implications for public health, particularly concerning obesity disorders. Integrating NLP-based indices, including sentiment scores patterns, into our predictive models significantly enhanced accuracy dyslipidemia predictions. This improvement was consistent across various machine learning techniques—XGBoost, LightGBM, CatBoost—demonstrating efficacy NLP methodologies refining disease prediction models. findings underscore critical role as indicators diseases. The successful application techniques offers a novel approach nutritional epidemiology, providing deeper understanding diet–disease nexus. contributes evolving field personalized nutrition emphasizes leveraging advanced computational tools inform targeted interventions strategies aimed at mitigating prevalence disorders population.

Language: Английский

Association between Prenatal, Pre-pregnancy Rainfall and Adult Obesity: Findings from the Community Behavior and Attitude Survey in Tuvalu (COMBAT) DOI Creative Commons
Chih‐Fu Wei,

Lois I. Tang,

Po-Jen Lin

et al.

Research Square (Research Square), Journal Year: 2025, Volume and Issue: unknown

Published: Feb. 10, 2025

Abstract Tuvalu has one of the highest obesity prevalence rates globally, and is a Pacific Island nation facing significant climate change challenges. Altered rainfall pattern, as part change, may influence risk during critical developmental periods. This study investigated associations between exposure prenatal, pre-pregnancy periods adult in Tuvalu. A nationwide survey was conducted February May 2022, which included 892 adults from Rainfall data obtained ECMWF Reanalysis v5 based on participants’ birth year birthplace. first birth, before two years analyzed, three to five were negative control Obesity severe defined body mass index (BMI) upon survey, according World Health Organization criteria. The results showed association higher increased BMI greater odds adulthood obesity. These more pronounced among male participants. No observed for birth. In conclusion, prenatal are associated with adulthood, reflecting environmental influences findings emphasize importance understanding climate-related health exposures need targeted interventions change-vulnerable populations. Further research should explore heterogeneity across nations mechanisms linking rainfall, weight,

Language: Английский

Citations

0
An Explainable Prediction for Dietary-Related Diseases via Language Models DOI Open Access
Insu Choi, Jihye Kim, Woo Chang Kim

et al.

Nutrients, Journal Year: 2024, Volume and Issue: 16(5), P. 686 - 686

Published: Feb. 28, 2024

Our study harnesses the power of natural language processing (NLP) to explore relationship between dietary patterns and metabolic health outcomes among Korean adults using data from Seventh Korea National Health Nutrition Examination Survey (KNHANES VII). Using Latent Dirichlet Allocation (LDA) analysis, we identified three distinct patterns: “Traditional Staple”, “Communal Festive”, “Westernized Convenience-Oriented”. These reflect diversity preferences in reveal cultural social dimensions influencing eating habits their potential implications for public health, particularly concerning obesity disorders. Integrating NLP-based indices, including sentiment scores patterns, into our predictive models significantly enhanced accuracy dyslipidemia predictions. This improvement was consistent across various machine learning techniques—XGBoost, LightGBM, CatBoost—demonstrating efficacy NLP methodologies refining disease prediction models. findings underscore critical role as indicators diseases. The successful application techniques offers a novel approach nutritional epidemiology, providing deeper understanding diet–disease nexus. contributes evolving field personalized nutrition emphasizes leveraging advanced computational tools inform targeted interventions strategies aimed at mitigating prevalence disorders population.

Language: Английский

Citations

0