Fairness-to-Action Framework Bridges AI Ethics Gap in Public Health Research

A new study from Sara Altamirano, Tijs Portegies, and Sennay Ghebreab, published in arXiv and accepted at IASEAI'26, tackles the persistent gap between awareness of algorithmic fairness and its actual implementation in public health machine learning. Using a sequential mixed-methods approach—expert interviews, an online survey, and systematic mapping—the researchers uncovered that fairness definitions remain fragmented across the field, training and guidance are limited, practitioners rely heavily on external sources, and formal assessment, mitigation, or monitoring of fairness is rare. These findings were mapped against three established gap frameworks: Knowledge-Practice Gap, Knowledge-to-Action Cycle, and Knowing-Doing Gap.

The paper introduces the 'Fairness-to-Action' framework, which integrates three complementary dimensions—methodological, organizational, and systemic—to identify where the translation of algorithmic fairness knowledge stalls. Analysis shows that fairness remains weakly institutionalized, translation mechanisms are externally driven, and system-level priorities still emphasize accuracy over fairness. The framework pinpoints critical leverage points for advancing safe, fair, and ethical ML-driven public health research, offering practical guidance for moving from awareness to action.