While the transition to artificial intelligence (AI) in the health and medical sectors is accelerating, an analysis has shown that South Korea's national university hospitals still lag behind in AI transformation in certain areas, such as access to personal health data, and face significant disparities among institutions.

The report titled "Analysis of Strengths, Weaknesses, and Disparities in Public Healthcare AX Based on Digital Health Maturity Diagnosis of National University Hospitals," published in the Korea Health Industry Development Institute's Bio-Health Industry Brief, yielded these findings based on data from 10 regional national university hospitals outside Seoul (including one public hospital) collected in 2025.

The evaluation consisted of 124 detailed items across four major categories: governance and workforce, interoperability, patient-centered healthcare, and predictive analytics. Among them, the governance and workforce category showed the highest performance with an AX achievement rate of 88.2%.

In contrast, patient-centered healthcare recorded the lowest achievement rate at 69.5%, with a standard deviation of 28.4 percentage points, indicating wider gaps among hospitals than in other categories.

In detail, "policies and decision-making," which focuses on data-driven strategic planning, and "interoperability structures," which encompass electronic prescriptions and the management and provision of medication history, were evaluated as common strengths across all 10 hospitals in terms of both AX achievement rates and standard deviations.

However, the "personalized" category—which includes access to health data and support for self-management linked with medical teams—showed an AX achievement rate of around 20.0% across the 10 hospitals, suggesting a structural vulnerability rather than an issue isolated to individual institutions.

The research team noted that while it was confirmed these national university hospitals possess substantial institutional and technological foundations for AX, such as governance and workforce capabilities, structural vulnerabilities were exposed in semantic data standardization, personal-level data linkage, and predictive analytics capabilities, which are essential for AI to function effectively.

They added that the low achievement rates in these areas reflect the reality of public healthcare where "data exists, but AI cannot read it." Since differences by region and scale were also confirmed among national university hospitals, the team suggested reviewing mentoring structures to connect top-tier and lower-tier institutions and establishing differentiated support policies based on capability levels for items with large disparities between organizations.

(Photo: Yonhap News)