Related Experiment Video
Updated: Sep 19, 2025
Assessment and Evaluation of the High Risk Neonate: The NICU Network Neurobehavioral Scale
Published on: August 25, 2014
Machine learning and explainable artificial intelligence to predict and interpret lead toxicity in pregnant women and
Priyanka Chaurasia1, Pratheepan Yogarajah1, Abbas Ali Mahdi2
1School of Computing, Engineering & Intelligent Systems, Ulster University, Londonderry, United Kingdom.
Insights
This study introduces explainable AI to interpret a machine learning model predicting lead exposure in pregnant women using sociodemographic data. This enhances trust and clinical applicability for maternal health interventions.
Area of Science:
- Environmental Health
- Artificial Intelligence
- Maternal Health
Background:
- Lead toxicity poses significant risks to infants, with prenatal exposure being a major concern.
- Maternal lead transfer during pregnancy necessitates accurate maternal lead level assessment for targeted interventions.
- Current lead detection methods are costly and inaccessible for widespread population screening.
Purpose of the Study:
- To enhance the transparency and interpretability of a previously developed machine learning (ML) model for predicting lead exposure in pregnant women.
- To build trust and facilitate informed decision-making in clinical and public health settings regarding lead exposure.
- To present the first application of an explainable artificial intelligence (XAI) framework for interpreting ML-based lead exposure predictions.
Main Methods:
- A Random Forest classifier was trained using a dataset of 200 blood samples and 12 sociodemographic features.
- Explainable Artificial Intelligence (XAI) methods, specifically SHAP (SHapley additive explanations) and LIME (Local Interpretable Model-Agnostic Explanations), were applied.
- These XAI methods were used to analyze feature contributions to the ML model's predictions.
Main Results:
- The Random Forest model achieved an accuracy of 84.52% in predicting lead exposure levels.
- SHAP and LIME provided insights into the contribution of each sociodemographic feature to the model's predictions.
- The application of XAI demonstrated how input features influence the prediction of maternal lead exposure.
Conclusions:
- Explainable AI frameworks can successfully interpret ML models used for predicting environmental health risks like lead exposure.
- Interpretable ML models are crucial for gaining acceptance and implementation in clinical practice and public health.
- This research paves the way for more transparent and trustworthy AI applications in maternal and child health protection.
Introduction:
Lead toxicity is a well-recognised environmental health issue, with prenatal exposure posing significant risks to infants. One major pathway of exposure to infants is maternal lead transfer during pregnancy. Therefore, accurately characterising maternal lead levels is critical for enabling targeted and personalised healthcare interventions. Current detection methods for lead poisoning are based on laboratory blood tests, which are not feasible for the screening of a wide population due to cost, accessibility, and logistical constraints. To address this limitation, our previous research proposed a novel machine learning (ML)-based model that predicts lead exposure levels in pregnant women using sociodemographic data alone. However, for such predictive models to gain broader acceptance, especially in clinical and public health settings, transparency and interpretability are essential.
Methods:
Understanding the reasoning behind the predictions of the model is crucial to building trust and facilitating informed decision-making. In this study, we present the first application of an explainable artificial intelligence (XAI) framework to interpret predictions made by our ML-based lead exposure model.
Results:
Using a dataset of 200 blood samples and 12 sociodemographic features, a Random Forest classifier was trained, achieving an accuracy of 84.52%.
Discussion:
We applied two widely used XAI methods, SHAP (SHapley additive explanations) and LIME (Local Interpretable Model-Agnostic Explanations), to provide insight into how each input feature contributed to the model's predictions.
