ARTIFICIAL INTELLIGENCE ETHICS AND PRINCIPLES IN PUBLIC SECTOR HEALTHCARE: A GOVERNANCE FRAMEWORK FOR RESPONSIBLE AI ADOPTION IN LOCAL AND SUB-NATIONAL HEALTH SERVICES

Povzetek

The proliferation of artificial intelligence (AI) and machine learning (ML) in public health systems presents profound ethical, regulatory, and governance challenges for local and sub-national governments. AI-enabled clinical decision support systems (CDSS) leveraging ensemble methods including Random Forest (RF) and Gradient Boosting Machines (GBM), unsupervised patient stratification techniques, and SHAP-based explainability offer significant potential for improving diagnostic accuracy, risk stratification, and service efficiency. However, deployment within publicly accountable healthcare institutions raises fundamental questions of transparency, fairness, accountability, human oversight, privacy, and robustness that performance metrics alone cannot resolve. This paper provides a comprehensive analysis of the ethical principles and governance frameworks required for responsible AI adoption in local public health services. Drawing on systematic literature review, comparative analysis of international regulatory instruments OECD AI Principles (2019), UNESCO Ethics Recommendation (2021), European Commission AI Regulation Proposal (2021), Council of Europe Convention 108+ (2018), GDPR Article 22 (2018), and the US AI Bill of Rights (2022) and a technical evaluation of ML approaches to CDSS, the paper proposes a five-component governance framework encompassing pre-deployment ethics assessment, transparency and explainability requirements, ongoing monitoring and accountability, public participation, and institutional capacity development. Implementation challenges including technical barriers, institutional capacity constraints, legal complexity, and structural equity asymmetries are examined in depth. Six policy recommendations are derived for statutory reform, procurement redesign, governance capacity, citizen rights, equity benchmarking, and evaluation research. The paper argues that trustworthy, equitable AI in local public health requires governance frameworks embedded in democratic institutions and informed by the communities whose health services are at stake.