Multimodal machine learning for analysing multifactorial causes of disease - The case of childhood overweight and obesity in Mexico

Aug 5, 2026

Mexico faces high rates of paediatric overweight and obesity, and public health interventions have achieved only moderate success. This study evaluates whether multimodal machine learning improves identification of predictive features from obesogenic environments and helps disentangle complex disease and social patterns using the Mexican National Health and Nutrition Survey (ENSANUT). Five feature modalities are tested in supervised and unsupervised multimodal pipelines and benchmarked against a unimodal early-fusion baseline.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages16

Reading Time40 min

AccessFree to download

Introduction

Paediatric obesity burden and relevance to Mexico

Purpose

Methods

Data modalities and multimodal learning pipelines

Supervised learning approach

Unsupervised learning approach

Results

Best-performing supervised model

Most important predictive features

Unsupervised clustering findings

Conclusion

TYPE Original Research PUBLISHED 07 January 2025 DOI 10. 3389/fpubh.2024.1369041 OPEN ACCESS EDITED BY Maria Oana Sasaran, “George Emil Palade” University of Medicine, Pharmacy, Sciences and Technology of Târgu Mures, Romania . REVIEWED BY Marco Bilucaglia, IULM University, Italy Martin Romero-Martínez, National Institute of Public Health, Mexico Álvaro Torres, University of Granada, Spain *CORRESPONDENCE Magnus Boman [magnus.boman@ki.se](magnus.boman@ki.se) RECEIVED 11 January 2024 ACCEPTED 16 December 2024 PUBLISHED 07 January 2025 CITATION Silva Sepulveda R and Boman M (2025) Multimodal machine learning for analysing multifactorial causes of disease—The case of childhood overweight and obesity in Mexico. Front. Public Health 12:1369041 . doi: 10.3389/fpubh.2024.1369041 COPYRIGHT © 2025 Silva Sepulveda and Boman. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY) . The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms. Multimodal machine learning for analysing multifactorial causes of disease—The case of childhood overweight and obesity in Mexico Rosario Silva Sepulveda1 and Magnus Boman1,2* 1 Karolinska Institutet, Department of Medicine Solna, Division of Clinical Epidemiology, Stockholm, Sweden, 2 MedTechLabs, BioClinicum, Karolinska University Hospital, Stockholm, Sweden Background: Mexico has one of the highest global incidences of paediatric overweight and obesity. Public health interventions have shown only moderate success, possibly from relying on knowledge extracted using limited types of statistical data analysis methods. Purpose: To explore if multimodal machine learning can enhance identifying predictive features from obesogenic environments and investigating complex disease or social patterns, using the Mexican National Health and Nutrition Survey. Methods: We grouped features into ﬁve data modalities corresponding topaediatric population exogenous factors, in two multimodal machine learning pipelines, against a unimodal early fusion baseline. The supervised pipeline employed four methods: Linear classiﬁer with Elastic Net regularisation, k-Nearest Neighbour, Decision Tree, and Random Forest. The unsupervised pipeline used traditional methods with k-Means and hierarchical clustering, with the optimal number of clusters calculated to be k = 2. Results: The decision tree classiﬁer in the supervised early fusion approach produced the best quantitative results. The top ﬁve most important features for classifying child or adolescent health were measures of an adult in the household, selected at random: BMI, obesity diagnosis, being single, seeking care at private healthcare, and having paid TV in the home. Unsupervised learning approaches varied in the optimal number of clusters but agreed on the importance of home environment features when analysing inter-cluster patterns. Main ﬁndings from this study di􀀀ered from previous studies using only traditional statistical methods on the same database. Notably, the BMI of a randomised adult within the household emerged as the most important feature, rather than maternal BMI, as reported in previous literature where unwanted cultural bias went undetected. Conclusion: Our general conclusion is that multimodal machine learning is a promising approach for comprehensively analysing obesogenic environments. The modalities allowed for a multimodal approach designed to critically analyse data signal strength and reveal sources of unwanted bias. In particular, it may aid in developing more e􀀀ective public health policies to address the ongoing paediatric obesity epidemic in Mexico. KE

FAQs

What is the main goal of this study on childhood overweight and obesity in Mexico?

The study explores whether multimodal machine learning can better identify predictive features from obesogenic environments and analyze complex disease or social patterns using ENSANUT data.

How did the researchers design the multimodal learning approaches?

Features were grouped into five data modalities and evaluated in two multimodal machine learning pipelines, compared with a unimodal early-fusion baseline. The supervised pipeline used Elastic Net, k-NN, decision trees, and random forest, while the unsupervised pipeline used k-means and hierarchical clustering with k=2.

Which features were most important for classifying child or adolescent health?

The top five most important features in the best supervised early-fusion approach were measures from an adult in the household, including BMI, obesity diagnosis, being single, using private healthcare, and having paid TV at home.

Aug 5, 2026

Upload by Theodore