AI Insight
This study quantitatively evaluated 39 Chinese national digital health policies from 2016-2025 using a Policy Modeling Consistency (PMC) framework with 48 indicators across nine dimensions. The analysis found a mean PMC score of 6.654 out of 10, with policies showing strong coverage in regulatory governance, standards, and data resources, but weaker and more variable coverage in data protection and industrial innovation. Nine policies were classified as perfect, 11 as good, 16 as acceptable, and three as poor based on structural completeness.
Why it matters
The findings reveal systematic gaps in China's digital health policy architecture, particularly in data protection and innovation support, which could inform more balanced policy development. The reproducible PMC framework offers other countries and regions a quantitative method to assess structural consistency in their own digital health policy systems.
Understand the Science
by Tong Lyu
Digital health governance increasingly relies on policy systems that coordinate data resources, information platforms, standards, service delivery, regulation, security, innovation, and implementation support. However, the structural composition and cross-policy heterogeneity of China’s national digital health policy system remain insufficiently quantified. This study develops a domain-specific Policy Modeling Consistency (PMC) framework to evaluate the structural coverage of national-level digital health management policies in China. Using a source-locked corpus covering 2016–2025, 42 eligible policies were identified, of which 39 with complete and auditable texts were included in the quantitative analysis. The framework comprises nine dimensions and 48 binary indicators, generating 1,872 policy-by-indicator coding decisions. An independently coded 11-policy subset yielded 96.78% agreement and a Cohen’s kappa of 0.914. The mean PMC score was 6.654 (SD = 1.456; median = 6.500; range = 3.467–9.000). Under the study’s structural-coverage thresholds, nine policies were classified as perfect, 11 as good, 16 as acceptable, and three as poor. Regulatory governance, standards and institutional arrangements, and data resources showed the highest average coverage, whereas data protection and industrial innovation showed the lowest coverage and the greatest cross-policy variation. Leave-one-policy-out and policy-family sensitivity analyses confirmed the stability of the principal structural pattern. These findings reveal a policy architecture with relatively extensive governance, institutional, and data-related provisions but more uneven protection- and innovation-related coverage. The proposed framework provides a reproducible basis for comparing policy-text structure while avoiding the interpretation of textual completeness as evidence of implementation quality or policy effectiveness.