Answers are grounded in this paper and cite their sources.
Learning Health Systems (LHSs) use routinely collected data to generate evidence that informs practice through continuous learning cycles. Artificial intelligence (AI) and data science may strengthen this model, but how LHS frameworks and AI applications relate remains unclear. We mapped LHS frameworks and their components and examined how AI and data science are integrated. Searches identified U.S.-based, English-language studies published January 2007–August 2025; two reviewers independently screened and extracted data, resolving disagreements by consensus. We included 54 framework studies (24 [44%] conceptual adaptations; 30 [56%] real-world implementations) and 30 AI/data science studies. Five components recurred across the frameworks: data infrastructure and interoperability, governance and stakeholder engagement, learning and improvement cycles, evidence generation and translation, and equity and ethics. Relative to framework studies, AI studies less often addressed governance and stakeholder engagement (17% vs 35%) or equity (13% vs 22%), emphasizing computational over organizational functions. Most AI studies (27/30) focused on the data-to-knowledge phase; only three extended to knowledge-to-practice, and none described practice-to-data feedback with post-deployment outcomes informing model updating. AI applications capture only part of the LHS vision, concentrating on knowledge generation with less attention to the organizational and feedback mechanisms needed to translate insights into practice and sustain learning.
No takes yet. Share an insight, caveat, or question.
Radwan et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: