Development and Validation of an Explainable Machine Learning Model to Assess the Prevalence Probability of Gastrointestinal Heat Retention Syndrome in Children: Cross-Sectional Study
Cross-sectional study develops an explainable model to assess gastrointestinal heat retention syndrome in children, aiding caregivers.
Key Points
The research aims to create and validate an interpretable machine learning model for assessing the prevalence probability of gastrointestinal heat retention syndrome in children.
Conducted a questionnaire survey with 120,198 kindergarten children in Longgang District, Shenzhen, China, excluding samples with missing GHRS information.
Utilized univariate logistic analysis, Least Absolute Shrinkage and Selection Operator regression, and Random Forest for model development and feature selection.
Built an online tool using the optimal machine learning model validated for performance and usability.
Collected 108,447 valid questionnaires, identifying 59 correlates of GHRS including 10 protective and 49 risk factors.
Random Forest model demonstrated high discriminatory performance and was selected as the primary analytical model after validation.
The online tool provides GHRS probability estimates and lifestyle recommendations based on responses to 75 questions.