Digital pathways connecting social and biological factors to health outcomes and equity

Abstract

Digital pathways extend conventional connections between social and biological factors and health outcomes, significantly influencing health equity. Data representation bias and distribution shifts are key mechanisms through which determinants of health impact generalizability of artificial intelligence (AI) models and subsequently affect health outcomes and equity. These mechanisms provide critical targets for algorithmic interventions, which can lead to Pareto improvements in AI model performance across diverse populations, thereby mitigating health disparities.