In an era marked by heightened public health emergencies and digital transformation, the capacity to communicate health risks transparently and equitably has never been more critical. Explainable Artificial Intelligence (XAI) offers novel pathways for enhancing health risk communication by enabling algorithmic decision-making processes to be understood by both experts and laypersons. This study investigates the effectiveness and equity implications of using XAI-driven tools in health risk communication frameworks. Employing a mixed-methods approach combining quantitative analysis of survey data and qualitative key informant interviews from three U.S. health departments, we assessed how XAI tools such as interpretable machine learning models influence public trust, risk comprehension, and behavioral intention across diverse populations. Findings indicate that while XAI-enhanced models improved overall risk comprehension by 31%, their benefits were significantly greater among participants with higher health literacy, raising concerns of a potential communication divide. Transparency, model explainability, and interactive features were positively associated with increased trust in health guidance. However, participants from marginalized communities expressed skepticism about data bias and fairness, emphasizing the need for culturally adapted explanations and participatory model design. Our results underscore that integrating explainable AI in public health communication must be accompanied by inclusive strategies to ensure equitable understanding and informed decision-making. Policy frameworks and technical standards for XAI deployment in health contexts must therefore prioritize transparency, inclusivity, and justice-centered design.
Blessing Agbaza-Mogbojuri (Wed,) studied this question.