Prediction Models for Early Identification of Overweight and Obese Children-A National Study
Irit Lior Sadaka1, Itamar Grotto2, Yair Sadaka3
1Department of Health Policy and Management, School of Public Health, Faculty of Health Sciences, Ben-Gurion University of the Negev, Be'er-Sheva 84105, Israel.
Insights
New machine learning models accurately predict childhood overweight risk using infant growth parameters. These models offer improved early detection compared to current World Health Organization (WHO) growth charts.
Area of Science:
- Pediatric endocrinology
- Machine learning in healthcare
- Public health and preventive medicine
Background:
- Early-life interventions are crucial for preventing childhood obesity.
- Accurate infant screening models are needed to identify high-risk individuals.
- Current World Health Organization (WHO) growth charts have limitations in predicting future overweight status.
Purpose of the Study:
- To develop and validate machine learning models for predicting childhood overweight.
- To utilize solely infant growth parameters for risk prediction.
- To outperform existing WHO growth chart predictions.
Main Methods:
- A retrospective national cohort study of infants born in Israel (2014-2016).
- Development of three machine learning models for age groups 0-3, 3-6, and 6-12 months.
- Comparison of model performance against WHO growth chart predictions using Area Under the Curve (AUC).
Main Results:
- Models demonstrated high predictive performance: AUCs of 0.76 (0-3m), 0.822 (3-6m), and 0.872 (6-12m).
- Models developed for 0-3 and 3-6 months showed superior prediction of early childhood overweight compared to WHO charts.
- A large cohort (198,503 children) ensured robust model validation.
Conclusions:
- Growth parameter-based machine learning models offer superior prediction of childhood overweight risk.
- These models are implementable globally in systems collecting infant growth data.
- A web calculator is available for practical risk assessment.
Background:
Given the importance of early-life intervention in reducing future obesity, screening models that more accurately identify infants at risk are crucial. The aim of this study was to develop models based solely on growth parameters for better predicting childhood overweight than the current WHO growth chart risk prediction.
Methods:
A retrospective national cohort study was conducted among children born in Israel between August 2014 and June 2016, followed for at least 18 months. Machine learning models were generated to predict childhood overweight. Three models for 0-3, 3-6, and 6-12 months were generated. These models were compared with the current WHO growth chart predictions. The outcome was defined as overweight in early childhood, based on weight-for-length (or weight-for-height) ≥97th percentile, according to WHO standards, measured at 18-36 months of age.
Results:
Overall, 198,503 children were included, and 150,572, 146,584, and 149,628 infants were included in Models 1, 2, and 3, respectively, with an average target age of two years. The models demonstrated high predictive performance (AUC) for the 0-3 months' model (0.76 [95% CI: 75 to 76.9%]), for the 3-6 months' model (0.822 [95% CI: 81.3 to 83.0%]) and for 6-12-month-old infants (0.872 [95% CI: 86.6 to 87.8%]). The first two models better predict the risk of early childhood overweight than the current WHO growth chart prediction.
Conclusions:
These models are unique in that they are based on growth parameters, usually screened at early childhood worldwide, and can be implemented in any system collecting growth measurements of infants, providing better risk prediction than the current WHO growth charts. A web calculator is provided.
More Related Videos
Related Concept Videos
01:24Obesity
01:26National Nursing Organizations I
01:30National Nursing Organizations II
The AACN emphasizes a healthy work environment through six standards to achieve an optimal patient outcome. The standards are appropriate staffing, meaningful recognition, collaboration, authentic leadership, effective communication, and decision-making. In addition, AACN provides certification programs, webinars, journals, and...
02:27Predicting Molecular Geometry
01:03Prediction Intervals
However, the point estimate is most likely not the exact value of the population parameter, but close to it. After calculating point estimates, we construct interval estimates, called confidence intervals or prediction intervals. This prediction interval comprises a range of values unlike the point estimate and is a better predictor of the observed sample value, y.
01:21Drug Dosing: Obese Patients