Related Experiment Video
Updated: Dec 8, 2025
A Pediatric Concussion Model in Mice: Closed Head Injury with Long-Term Disorders (CHILD)
Published on: February 7, 2025
Validation of a Machine Learning Model to Predict Childhood Lead Poisoning
Eric Potash1, Rayid Ghani2, Joe Walsh1
1Harris School of Public Policy, University of Chicago, Chicago, Illinois.
Insights
A new machine learning model effectively predicts childhood lead poisoning, outperforming traditional regression methods. This advancement enables targeted prevention resources for high-risk children, mitigating irreversible neurodevelopmental harm.
Area of Science:
- Environmental Health
- Pediatrics
- Data Science
Background:
- Childhood lead poisoning results in permanent neurobehavioral deficits.
- Current prevention strategies primarily focus on secondary prevention after exposure.
- Early identification of children at risk is crucial for effective intervention.
Purpose of the Study:
- To validate a machine learning (random forest) prediction model for elevated blood lead levels (EBLLs).
- To compare the performance of the random forest model against a logistic regression model.
- To assess the utility of predictive models for targeting lead poisoning prevention efforts.
Main Methods:
- A prognostic study utilized data from the Chicago Department of Public Health's WIC program.
- A development cohort (2007-2012) and a validation cohort (2013) were established.
- Blood lead levels were measured, and models were evaluated using AUC and confusion matrix metrics.
Main Results:
- The random forest model achieved a higher AUC (0.69) compared to logistic regression (0.64).
- For predicting the highest-risk 5% of children, random forest showed improved positive predictive value (15.5% vs. 7.8%) and sensitivity (16.2% vs. 8.1%).
- Both models demonstrated high specificity, with random forest at 95.5% and logistic regression at 95.1%.
Conclusions:
- Machine learning models, specifically random forest, demonstrate superior performance in predicting childhood lead poisoning compared to logistic regression.
- The enhanced predictive accuracy, particularly in identifying high-risk children, supports the use of machine learning for resource allocation.
- Implementing such models can optimize the targeting of lead poisoning prevention resources, improving public health outcomes.
Importance:
Childhood lead poisoning causes irreversible neurobehavioral deficits, but current practice is secondary prevention.
Objective:
To validate a machine learning (random forest) prediction model of elevated blood lead levels (EBLLs) by comparison with a parsimonious logistic regression.
Design, Setting, And Participants:
This prognostic study for temporal validation of multivariable prediction models used data from the Women, Infants, and Children (WIC) program of the Chicago Department of Public Health. Participants included a development cohort of children born from January 1, 2007, to December 31, 2012, and a validation WIC cohort born from January 1 to December 31, 2013. Blood lead levels were measured until December 31, 2018. Data were analyzed from January 1 to October 31, 2019.
Exposures:
Blood lead level test results; lead investigation findings; housing characteristics, permits, and violations; and demographic variables.
Main Outcomes And Measures:
Incident EBLL (≥6 μg/dL). Models were assessed using the area under the receiver operating characteristic curve (AUC) and confusion matrix metrics (positive predictive value, sensitivity, and specificity) at various thresholds.
Results:
Among 6812 children in the WIC validation cohort, 3451 (50.7%) were female, 3057 (44.9%) were Hispanic, 2804 (41.2%) were non-Hispanic Black, 458 (6.7%) were non-Hispanic White, and 442 (6.5%) were Asian (mean [SD] age, 5.5 [0.3] years). The median year of housing construction was 1919 (interquartile range, 1903-1948). Random forest AUC was 0.69 compared with 0.64 for logistic regression (difference, 0.05; 95% CI, 0.02-0.08). When predicting the 5% of children at highest risk to have EBLLs, random forest and logistic regression models had positive predictive values of 15.5% and 7.8%, respectively (difference, 7.7%; 95% CI, 3.7%-11.3%), sensitivity of 16.2% and 8.1%, respectively (difference, 8.1%; 95% CI, 3.9%-11.7%), and specificity of 95.5% and 95.1% (difference, 0.4%; 95% CI, 0.0%-0.7%).
Conclusions And Relevance:
The machine learning model outperformed regression in predicting childhood lead poisoning, especially in identifying children at highest risk. Such a model could be used to target the allocation of lead poisoning prevention resources to these children.
