Shaping the future of AI in healthcare through ethics and governance

The integration of artificial intelligence (AI) into healthcare presents both immense opportunities and significant challenges. AI systems promise to revolutionize healthcare by improving diagnostic accuracy, enhancing treatment efficacy, accelerating drug discovery, and optimizing care management. However, the rapid advancement of AI also raises critical ethical, legal, and social concerns. Data privacy, algorithmic bias, patient safety, and the potential for exacerbating existing health inequities are key issues demanding careful consideration. The World Health Organization (WHO) has acknowledged these challenges and issued reports outlining principles for the ethical and responsible use of AI in healthcare. These reports emphasize equity, human dignity, and fundamental rights, but they lack the legal force to compel action. This necessitates a stronger international legal framework to address the complexities and vulnerabilities inherent in AI's deployment within healthcare systems. The European Union's (EU) comprehensive legal framework, including the General Data Protection Regulation (GDPR), the Data Act, and the Artificial Intelligence (AI) Act, serves as a potential model for global harmonization, offering a more robust approach to AI governance than currently exists at the international level. This research examines the existing gaps in AI regulation, analyzes the potential of EU regulations as a model, and proposes a path forward for creating a global legal framework to guide the ethical and responsible use of AI in healthcare.

The literature review examines existing research on the use of AI in healthcare, encompassing a broad range of applications, from care management and medical imaging analysis to drug discovery and precision medicine. It also analyzes the ethical and regulatory challenges associated with AI implementation, including concerns about data privacy, algorithmic bias, and health equity. The review considers existing guidelines and recommendations from the WHO and the EU, highlighting the current gaps in harmonization and the need for legally binding international standards. Studies on the impact of AI on various aspects of healthcare, including diagnostic accuracy, treatment effectiveness, and patient outcomes, are also incorporated. Finally, the insufficiency of existing legal frameworks and their inability to keep pace with rapid technological advancements are examined.

This research is based on a comprehensive literature review of publicly available data up to February 2024. The initial screening focused on articles' titles and abstracts, followed by a full-text evaluation of eligible articles. Searches were conducted using Google Scholar, focusing on AI applications in healthcare, with attention given to applications not explicitly mentioned by the EU or WHO. The search also included specific applications of AI in healthcare to understand how health systems currently manage AI-related challenges. The WHO's institutional repositories were also explored. The research assessed the adequacy of current legal frameworks, particularly by examining the differences between the binding regulations of the EU and the non-binding guidelines of the WHO. This highlights the need for clearer international standards, suggesting the importance of prioritizing international and global laws over national regulations. The analysis considers the lack of universal agreement on AI use in healthcare.

The research reveals a significant gap in the global legal framework regulating AI in healthcare. While the WHO has provided guiding principles, these are non-binding recommendations. The EU's comprehensive approach, including the GDPR, Data Act, and AI Act, offers a more robust model. The study identifies key challenges associated with AI in healthcare, including: data privacy and security; algorithmic bias and the potential for discrimination; health equity and access; transparency and explainability of AI systems; and the need for clear accountability mechanisms. The research emphasizes the importance of harmonized international standards to ensure safe and ethical AI development and deployment. Specific examples of AI applications in healthcare are analyzed, including their potential benefits and associated risks. The need for comprehensive risk assessment and mitigation strategies is highlighted, along with the importance of effective regulatory oversight. The EU’s AI Act categorizes AI systems based on risk levels, implementing stringent rules for high-risk systems, and transparency requirements for general-purpose AI. The WHO's guiding principles focus on protecting autonomy, promoting safety, ensuring transparency, fostering responsibility, ensuring equity, and promoting sustainable AI development. The study argues for a legally binding international framework, possibly amending the International Health Regulations (IHR), to address the global implications of AI in healthcare, emphasizing the duty of states to cooperate under the UN Charter in promoting global health and safeguarding fundamental human rights.

The findings highlight the critical need for a comprehensive and harmonized international legal framework for AI in healthcare. The current landscape of fragmented national regulations and non-binding guidelines leaves significant gaps in governance, risking both the misuse of AI and the inability to fully realize its potential benefits. The EU's regulatory approach provides a strong foundation for potential global adoption, demonstrating the feasibility of establishing legally binding standards while also promoting innovation. The ethical challenges related to data privacy, algorithmic bias, and health equity demand solutions that balance innovation with the protection of human rights. International cooperation, driven by organizations such as the WHO, is crucial for building consensus and establishing universally accepted standards. The WHO's constitutional mandate and experience in coordinating global health responses position it effectively to guide this process. The proposed amendment of the IHR, reinforced by the UN Charter's duty of cooperation, could provide the necessary legally binding framework.

The study concludes that a global legal framework for AI in healthcare is urgently needed to address the ethical, legal, and technical challenges. The WHO should take a leading role in developing legally binding rules, possibly through amendments to the IHR, drawing inspiration from the EU's comprehensive regulatory approach. International cooperation and collaboration among stakeholders are crucial for successful implementation. Future research could focus on specific AI applications, developing detailed risk assessment methodologies, and exploring innovative governance models to promote both responsible AI development and equitable access to healthcare.

The study's reliance on publicly available data up to February 2024 limits its scope. Rapid advancements in AI may necessitate ongoing updates. The focus on the EU regulatory framework as a model may not fully reflect the diverse needs and contexts of all nations. Further research is needed to fully explore the practical implications of implementing a global regulatory framework.