A Hybrid Approach to Modelling ECC Risk: Effectiveness of Nonparametric Regression and MLFNN Techniques
Introduction. Early childhood caries (ECC) is a prevalent dental condition that significantly impacts children’s quality of life and is influenced by both environmental and metabolic factors. Anthropometric variables such as weight, height, and body mass index serve as indicators of overall growth a...
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Ivano-Frankivsk National Medical University
2025-03-01
|
| Series: | Galician Medical Journal |
| Subjects: | |
| Online Access: | https://ifnmujournal.com/gmj/article/view/2058 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Introduction. Early childhood caries (ECC) is a prevalent dental condition that significantly impacts children’s quality of life and is influenced by both environmental and metabolic factors. Anthropometric variables such as weight, height, and body mass index serve as indicators of overall growth and nutritional status, which are closely linked to oral health. Additionally, family size may play a role in ECC risk by influencing dietary habits and oral hygiene practices. Despite these associations, the complex and nonlinear relationships between these factors and ECC risk remain insufficiently explored.
This study aims to investigate the potential relationships between ECC risk and a combination of social (family size) and anthropometric parameters (weight, height, and body mass index) by employing nonparametric regression and validating these relationships with a multilayer feed-forward neural network (MLFNN).
Methods. This cross-sectional observational study utilized secondary data from Universiti Sains Malaysia Hospital, Kota Bharu, Kelantan, Malaysia. The dataset was divided into training (60%), testing (30%), and validation (10%) subsets. A generalized additive model (GAM) was used to capture nonlinear relationships, followed by MLFNN validation. Model performance was assessed using root mean squared error (RMSE), mean absolute error (MAE), median absolute error (MedAE), and mean squared error (MSE).
Results. The dataset exhibited non-normality, justifying the use of nonparametric regression. In the GAM, only weight showed a nonlinear relationship, with no other predictors significantly influencing ECC risk, except for the intercept. The MLFNN achieved an error value of 0.089 and an accuracy of 91.09%, with height contributing the most to ECC risk estimation.
Conclusions. Integrating nonparametric regression and MLFNN validation provides a robust framework for modelling ECC risk, capturing complex nonlinear relationships between family size and anthropometric factors. Height emerged as the most influential predictor, highlighting its association with growth and systemic health, followed by weight, body mass index, and family size. These findings underscore the need for a multifactorial approach in ECC prevention, emphasizing nutritional and family-related factors in pediatric dental care. |
|---|---|
| ISSN: | 2414-1518 |