aceIdentification of obesity risk factors in 3-12-year-old children and adolescents with prior respiratory tract infections via interpretable machine and deep learning models
Abstract: Childhood obesity and respiratory tract infections (RTIs) are 2 major global public health issues that frequently co-occur and are closely interrelated. Early detection of children with prior RTIs who are at high obesity risk is crucial for targeted interventions. This study integrates interpretable machine learning (ML) models and a deep learning network to develop an obesity risk prediction model in a large pediatric cohort. Cross-sectional data from 6509 children and adolescents aged 3-12 years with prior RTI…
High-Priority Gaps for Clinical Preventive Services: Seventh Annual Report To Congress 2017 by U.S. Preventive Services Task Force (USPSTF). Public domain
In a study of 6509 children and adolescents aged 3-12 with prior respiratory tract infections in Beijing and Tangshan, researchers developed and compared 12 machine learning models to predict obesity risk, with LightGBM achieving the best reported performance and a deep learning sequence network used to verify the selected features.
The work matters because it links interpretable AI to early identification of high-risk children where obesity and respiratory infections co-occur, offering a tool for targeted screening, while uncertainty remains about generalizability beyond the two Chinese cities, cross-sectional design, and real-world implementation.
- Study used cross-sectional data from 6509 children and adolescents aged 3-12 years with prior RTIs in Beijing and Tangshan across 12 ML models.
- Bayesian optimization was applied to fine-tune model hyperparameters and performance assessed using 8 metrics.
- SHAP analysis identified 20 key predictors including child age, paternal BMI, maternal BMI, birth length, birthweight, gestational age, eating speed, and screen time.
- Deep learning sequential neural network validated the feature set with accuracy 0.8023 and AUC 0.8117, with SHAP rankings in close line with LightGBM.
LightGBM-based model predicted obesity versus normal weight in children aged 3-12 with prior RTIs with high accuracy and AUC, enabling early screening for targeted intervention.
The rundown
Researchers fed cross-sectional data from 6509 children and adolescents in Beijing and Tangshan to 12 machine learning models to distinguish obesity from normal weight among those with prior respiratory tract infections, using Bayesian optimization for hyperparameter tuning and 8 metrics for evaluation.
LightGBM was selected as optimal and interpreted with SHAP, which highlighted 20 predictors spanning parental BMI, birth characteristics, eating speed, screen time, sleep, and family history, with a sequential neural network providing additional validation of predictive value.
Sources
- Peer-reviewedJAMIA Open2026-08-06
How should this claim be treated?
The debate