Related Experiment Video
Updated: Oct 22, 2025
12:18
A Machine Learning Approach to Design an Efficient Selective Screening of Mild Cognitive Impairment
Published on: January 11, 2020
7.7K
Prediction Models of Early Childhood Caries Based on Machine Learning Algorithms
You-Hyun Park1, Sung-Hwa Kim1, Yoon-Young Choi2
1Department of Biostatistics, Yonsei University Wonju College of Medicine, Wonju 26426, Korea.
Summary
Machine learning models and traditional logistic regression effectively predict early childhood caries (ECC) risk in young children. Both approaches show similar performance, aiding in identifying high-risk groups for preventive care.
Area of Science:
- Oral Health
- Data Science
- Public Health
Background:
- Early childhood caries (ECC) poses a significant public health challenge.
- Accurate prediction models are crucial for timely intervention and prevention strategies.
Purpose of the Study:
- To develop and compare machine learning (ML)-based prediction models for ECC with traditional logistic regression.
- To evaluate the performance of XGBoost, random forest, and LightGBM algorithms against logistic regression for ECC prediction.
Main Methods:
- Analysis of 4195 children aged 1-5 years from the Korea National Health and Nutrition Examination Survey (2007-2018).
- Development of prediction models using logistic regression, XGBoost, random forest, and LightGBM.
- Variable selection using regression-based backward elimination and random forest-based permutation importance.
Main Results:
- All four models demonstrated comparable predictive performance, with Area Under the Receiver Operating Characteristic (AUROC) values ranging from 0.774 to 0.785.
- No statistically significant differences were found between the AUROC values of the logistic regression and ML-based models.
Conclusions:
- Both traditional logistic regression and ML-based models are suitable for predicting ECC and identifying high-risk children.
- These models can support the implementation of targeted preventive treatments for early childhood caries.
- Further research incorporating advanced methods like deep learning is recommended to enhance prediction model performance.
