Using agent-based models and EXplainable Artificial Intelligence (XAI) to simulate social behaviors and policy intervention scenarios: A case study of private well users in Ireland

Free testing removes the primary barrier to well safety: The study found that offering free well testing alone increased participation from ~5% to over 75%, demonstrating that cost is the dominant factor preventing Irish households from protecting their water supply. This suggests that financial incentives may be more effective than education or regulation alone.

Combining free testing with communication amplifies public health impact: The most successful policy scenario (Free Well Testing plus Communication Campaign) achieved 435 out of 561 agents (77%) participating in well testing, compared to baseline rates near 5%. This indicates that pairing financial support with targeted messaging creates stronger behavioral change than either intervention alone.

AI can make policy recommendations more transparent and trustworthy: The researchers used Explainable AI (SHAP) to identify which factors most influence well-testing decisions, moving beyond "black box" predictions. This approach allows policymakers to understand why the AI recommends certain interventions, supporting more accountable decision-making in public health policy.

Agent-based modeling enables testing policies before implementation: Rather than rolling out expensive or disruptive policies without evidence, this study simulated 14 different scenarios using real survey data. This allows governments to predict outcomes and choose the most cost-effective approach—reducing wasted resources and protecting public health more efficiently.

Behavioral adaptation happens quickly when barriers are removed: The study found "rapid behavioral adaptation" when free testing was offered, suggesting that people want to protect their families' health but face practical obstacles. Removing these obstacles produces immediate, measurable improvements in safety-seeking behavior.