Governance, not technology: low-hanging fruits for AI readiness in Europe, a latent class analysis
Abstract
Artificial intelligence adoption varies widely across the WHO European Region, yet governance maturity remains poorly characterised. This exploratory study applies latent class analysis to identify distinct digtal governance maturity profiles among Member States and to determine statistically which governance factors most predict higher AI readiness, offering countries an evidence base for actionable low-hanging fruits. Data from two WHO surveys, the Assessment of AI Readiness and the Digital Health Country Survey, were harmonised across 50 Member States. Indicators spanning five WHO governance pillars were selected through reliability testing. Latent class models with two to five classes were estimated; model selection balanced information criteria, entropy, and interpretability. Findings are exploratory pending further validation. Four preliminary profiles emerged. Infrastructure investment was near-universal yet insufficient to differentiate maturity levels. The factors clearly separating lower from higher AI readiness countries were ethics and accountability frameworks, national AI strategies, and workforce readiness. These factors were absent in lower-maturity profiles despite comparable infrastructure, confirming that AI readiness is a governance issue, not technological. Countries can use their profile to identify precise governance gaps rather than investing broadly across all pillars. Ethics and accountability frameworks showed the largest gap between developing and mature profiles and represent tractable low-hanging fruit. Linking regulatory capacity to accountability structures and formalising national AI strategies are high-leverage, low-cost actions. Technology infrastructure did not differentiate mature from developing profiles. These findings suggest governance maturity is configurational rather than linear. Profile-based approaches can help WHO and Member States move from uniform capacity-building to targeted, context-specific policy support.
Key messages: • Countries do not need more technology to improve AI readiness. Ethics frameworks, national AI strategies and accountability structures are the low-hanging fruits that most predict governance maturity.
• Knowing your governance profile tells you exactly where to act. Countries with similar AI readiness scores often need opposite fixes that composite indices and uniform policy guidance would miss.
Related
- Recent biotechnological advances and technology readiness level of natural products derived from marine fungi
- Artificial intelligence for Traditional Chinese Medicine classic formula development: a technology readiness assessment using metabolic syndrome as a case study
- AI in health as a governance challenge: findings from WHO Europe’s knowledge community
- A 25‐Year Nationwide Analysis of Mortality Trends due to Intentional Self‐Harm by Hanging, Strangulation, and Suffocation
- Hanging on through Omicron, then what? A pre-exit baseline of the U.S. emergency nursing workforce, 2018 to 2022, with implications for the 2026 NSSRN cycle
BibTeX
@article{sy2026governance,
title = {{Governance, not technology: low-hanging fruits for AI readiness in Europe, a latent class analysis}},
author = {TRL Sy and ES Baja},
journal = {The European Journal of Public Health},
year = {2026},
month = oct,
doi = {10.1093/eurpub/ckag169.883},
url = {https://doi.org/10.1093/eurpub/ckag169.883},
note = {CC BY-NC 4.0},
}
RIS
TY - JOUR
TI - Governance, not technology: low-hanging fruits for AI readiness in Europe, a latent class analysis
AU - TRL Sy
AU - ES Baja
PY - 2026
DA - 2026/10/01
JO - The European Journal of Public Health
DO - 10.1093/eurpub/ckag169.883
UR - https://doi.org/10.1093/eurpub/ckag169.883
AB - Artificial intelligence adoption varies widely across the WHO European Region, yet governance maturity remains poorly characterised. This exploratory study applies latent class analysis to identify distinct digtal governance maturity profiles among Member States and to determine statistically which governance factors most predict higher AI readiness, offering countries an evidence base for actionable low-hanging fruits. Data from two WHO surveys, the Assessment of AI Readiness and the Digital Health Country Survey, were harmonised across 50 Member States. Indicators spanning five WHO governance pillars were selected through reliability testing. Latent class models with two to five classes were estimated; model selection balanced information criteria, entropy, and interpretability. Findings are exploratory pending further validation. Four preliminary profiles emerged. Infrastructure investment was near-universal yet insufficient to differentiate maturity levels. The factors clearly separating lower from higher AI readiness countries were ethics and accountability frameworks, national AI strategies, and workforce readiness. These factors were absent in lower-maturity profiles despite comparable infrastructure, confirming that AI readiness is a governance issue, not technological. Countries can use their profile to identify precise governance gaps rather than investing broadly across all pillars. Ethics and accountability frameworks showed the largest gap between developing and mature profiles and represent tractable low-hanging fruit. Linking regulatory capacity to accountability structures and formalising national AI strategies are high-leverage, low-cost actions. Technology infrastructure did not differentiate mature from developing profiles. These findings suggest governance maturity is configurational rather than linear. Profile-based approaches can help WHO and Member States move from uniform capacity-building to targeted, context-specific policy support. Key messages: • Countries do not need more technology to improve AI readiness. Ethics frameworks, national AI strategies and accountability structures are the low-hanging fruits that most predict governance maturity. • Knowing your governance profile tells you exactly where to act. Countries with similar AI readiness scores often need opposite fixes that composite indices and uniform policy guidance would miss.
SN - 1101-1262
SN - 1464-360X
ER -