Objective:

To develop and evaluate a machine learning framework for predicting the likelihood of gastrointestinal heat retention syndrome (GHRS) in pediatric patients.

Approach:

• Model Development: The diagnostic model for GHRS was developed using the Extreme Gradient Boosting (XGBoost) algorithm, achieving an internal validation accuracy of 93.03%.
• Cross-Sectional Study: A cross-sectional study was conducted in Beijing to assess the prevalence of GHRS among children, analyzing sociodemographic factors and lifestyle habits.

Key Findings:

• GHRS is characterized by gastrointestinal symptoms such as dry stool and straining during defecation.
• Higher risk factors for GHRS include age 3-5 years, being female, urban residency, and certain dietary habits.
• Protective factors against GHRS include daily indoor exercise and higher consumption of vegetables, fish, and soy products.

Interpretation:

The study identifies significant sociodemographic and lifestyle factors associated with GHRS, highlighting the need for further research on its prevalence and clinical implications.

Limitations:

• Limited research on GHRS prevalence prediction models.
• The study's findings are based on a specific population in Beijing, which may not be generalizable.

Conclusion:

The machine learning framework developed for GHRS prediction shows promise, but further validation and exploration of its clinical utility are necessary.