Unfolding the policy dynamics of medical data assetization in Chinese public healthcare institutions: evidence from Latent Dirichlet Allocation and dynamic topic modeling analysis.

Background: As medical data increasingly functions as a factor of production, public healthcare systems face the challenge of reconciling data-driven innovation with regulatory accountability. In China, a series of national initiatives has sought to promote the assetization of medical data through market-oriented mechanisms. However, existing studies have primarily focused on legal frameworks, technical architectures, and ownership debates, while the temporal evolution of policy priorities and the interaction between central design and local implementation remain insufficiently understood. Methods: This study constructs a corpus of 192 policy documents (47 national and 145 provincial) issued between 2015 and 2025 and applies Latent Dirichlet Allocation (LDA) and Dynamic Topic Modeling (DTM) to examine both the structural composition and temporal evolution of policy agendas. Specifically, LDA is used to identify latent thematic structures at national and provincial levels, while DTM traces shifts in topic intensity across policy phases and captures the dynamic relationship between central agenda-setting and provincial policy activation. Results: The results reveal a clear differentiation in governance focus. National policies emphasize macro-level institutional design, security governance, and authorized operation, with valuation-related discourse exhibiting a short-term surge followed by a strategic reorientation toward operational feasibility. Provincial policies, by contrast, display a more implementation-oriented trajectory, progressively prioritizing registration procedures, circulation mechanisms, and trading infrastructure. Dynamic analysis further shows that policy diffusion follows a staged pattern rather than linear synchronization, with 2023 marking a critical inflection point for local policy activation. Conclusion: Together, the findings indicate a functional division of policy labor between central and local governments. Medical data assetization in China is advancing through a cautious and adaptive governance process in which authorization, registration, and secure circulation precede stable valuation. This study provides empirical evidence on the sequencing logic of medical data assetization and demonstrates the value of computational policy analysis for understanding multi-level governance dynamics in public health data governance.