Answers are grounded in this paper and cite their sources.
Background Cities have increasingly invested in public health data dashboards and related interfaces to address challenges such as opioid use, maternal health outcomes, and health inequities. While these tools improve transparency, prior work has shown that such systems rarely support continuous learning, cross‐sector coordination, or adaptive action during implementation. In many settings, fragmented data ownership, distributed authority, and limited mechanisms for iteration constrain the translation of data into sustained public health improvement. Approach This article extends learning health system theory by conceptualizing the city itself as the learning boundary for public health action. We describe a model in which learning is organized around public health initiatives spanning multiple organizations and supported by a city‐level health data enclave built on a shared, deidentified data architecture for cross‐organizational learning. Central to this approach are governance structures that foster trust, align priorities, and sustain learning across independent organizations. Implementation Experience Drawing on direct experience developing the Worcester Integrated Health Data Exchange (WIHDE), an ongoing city‐academic‐healthcare initiative in Worcester, Massachusetts, we describe how initiative‐driven learning can reorient data use from retrospective reporting toward action‐oriented public health practice. Worcester provides a practical example grounded in ongoing implementation, illustrating how shared infrastructure and governance are being established to support this approach. Conclusions By reframing the learning boundary from institutions to cities, this article offers a generalizable framework for organizing continuous, action‐oriented learning in complex public health systems.
No takes yet. Share an insight, caveat, or question.
Zai et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: