Advancing health equity in proactive health management: from data underrepresentation and algorithmic bias to a closed-loop governance framework

Abstract

Proactive health management (PHM) relies on wearable devices, mobile health applications, and artificial intelligence (AI) to continuously monitor the health status of individuals, thus supporting risk stratification and early intervention. The effectiveness of such techniques fundamentally depends on whether training data adequately represents the target population. Structural data gaps in patient-generated health data (PGHD), driven by social determinants of health (SDoH) and the digital divide, result in the systematic underrepresentation of high-risk groups in training data. This underrepresentation cascades through model development and deployment, leading to prediction bias, risk omission in vulnerable populations, and misallocation of health resources. To synthesize current evidence on this topic, we conducted a narrative review searching PubMed, Web of Science, and Scopus for peer-reviewed literature published between January 2015 and February 2026, using terms related to proactive health management, artificial intelligence, algorithmic bias, and health equity. A total of 108 studies meeting predefined inclusion criteria were included and synthesized thematically. This review analyzes the mechanisms driving insufficient data representativeness in PHM and traces how prediction bias accumulates across model development and deployment, resulting in health inequities. We identify that equity risk in PHM stems from the interplay of three factors: social inequality structures, data collection mechanisms, and predictive target design. Because these drivers operate across the entire PHM lifecycle, addressing equity requires integrated, system-wide governance rather than isolated technical adjustments. This paper therefore proposes a comprehensive closed-loop governance framework encompassing targeted data collection, fairness constraint mechanisms, pre-deployment subgroup audits, and post-deployment continuous monitoring. By transforming fairness from a technical condition into a core systemic principle, this framework aims to ensure health equity throughout the PHM lifecycle.