China’s national digital-health policies commonly address governance, standards and data, but provisions for data protection and industrial innovation appear less consistently across policy texts. That is the main pattern in a study by Tong Lyu, published in PLOS ONE, which compares what the documents explicitly cover.
Lyu identified 42 eligible national policies issued between January 2016 and December 1, 2025. The quantitative analysis included 39 whose complete texts could be archived in an auditable form. The study used a framework designed for digital-health policy to assess structural coverage: whether specified elements appeared in each text. It does not measure whether policies were implemented or improved health services.
The framework divided coverage into nine dimensions, including data resources, digital platforms, service delivery, governance, data protection and innovation. Across those dimensions, the study applied 48 binary indicators to each policy, producing 1,872 coding decisions. A second human researcher independently coded 11 policies; the coders agreed on 96.78% of decisions, with a Cohen’s kappa of 0.914. That reliability check applies to the 11-policy subset, not the entire corpus, and the study notes that coding still involves judgment.
The policies’ average score was 6.654 out of a possible nine. Scores ranged from 3.467 to 9.000. The researchers classified policies into four categories—poor, acceptable, good and perfect—using thresholds set for this framework. Those labels describe textual coverage only. A lower score may reflect a policy’s narrower purpose, rather than inferior design or weak implementation.
Regulatory governance had the highest average coverage, followed by standards and institutional arrangements, then data resources. Data-enabled applications appeared in all 39 analyzed texts. But policies did not consistently combine provisions on data catalogs, authorized access, sharing, data quality, classification and lifecycle governance. The study therefore distinguishes frequent references to using health data from the more variable arrangements for managing it.
Data protection had the lowest average coverage, while industrial innovation had the second-lowest. Both dimensions also varied considerably among policies. Protection indicators included cybersecurity responsibilities, personal-information safeguards and security monitoring. Innovation indicators included technology development, pilot activities and collaboration. The study’s findings show that these topics are unevenly represented in the documents; they do not establish that health data are insecure in practice or that innovation activity is limited.
The pattern held up in checks that recalculated results while leaving out one policy at a time and while combining a closely related policy family. Even so, the analysis is a map of policy texts, not a test of real-world effects. It cannot show whether provisions were enforced, adequately resourced or associated with changes in service quality or health outcomes. The study also excluded three eligible policies because their complete texts could not be archived or reliably converted for analysis.
The framework offers a structured way to compare which elements are stated across national digital-health policies and where coverage differs. Its value, the study suggests, is in making those textual patterns visible while keeping them separate from questions of implementation and effectiveness.
<a href="https://journals.plos.org/plosone/article?id=10.1371%2Fjournal.pone.0362220">Source: PLOS ONE article</a>