A prediction model for childhood obesity risk using the machine learning method: a panel study on Korean children

Area of Science:

• Pediatrics
• Public Health
• Machine Learning in Healthcare

Background:

• Children face an obesogenic environment with processed foods and reduced activity.
• Maternal perceptions and parenting styles impact children's weight management.
• Childhood obesity is a growing public health concern requiring predictive models.

Purpose of the Study:

• To develop a machine learning-based prediction model for childhood obesity in 10-year-olds.
• To identify significant risk factors contributing to childhood obesity.
• To explore the relationship between maternal factors and childhood obesity.

Main Methods:

• Utilized data from 1185 children and mothers from the Korean National Panel Study.
• Developed a prediction model using the least absolute shrinkage and selection operator (LASSO) method.
• Included ten factors: child's gender, eating habits, activity, BMI, and mother's education, self-esteem, BMI.

Main Results:

• The prediction model achieved an Area Under the Receiver Operator Characteristic Curve of 0.82 and 76% accuracy.
• Significant risk factors identified: child's physical activity levels and maternal self-esteem.
• Maternal self-esteem emerged as a novel predictor of childhood body mass index.

Conclusions:

• Maternal self-esteem is significantly associated with childhood obesity risk.
• The developed model effectively predicts childhood obesity.
• Further research is needed for targeted interventions for at-risk children and mothers.