WHO urges enhanced ethical oversight for AI health research

The World Health Organization has emphasised the necessity for enhanced and more unified ethical oversight concerning artificial intelligence-related health research, cautioning that current frameworks may lack the capacity to effectively manage the risks arising from the swift adoption of AI in healthcare research. In a recent report released on Monday, the WHO indicated that AI is accelerating research and reducing the time required to transition from experimentation to practical applications. However, it also highlighted significant concerns regarding privacy, bias, fairness, transparency, accountability, and the swift implementation of AI-driven tools. According to the WHO, while AI has the potential to generate new knowledge and enhance health outcomes, it also introduces risks that diverge from those encountered in traditional health research.

One concern is speed. The research-to-product cycle for AI tends to be significantly shorter than that of traditional medical technologies. This brevity implies that issues not detected during the research phase can rapidly manifest in products utilised by patients or health systems. AI research may yield results that are effective in a specific population or healthcare context, yet these findings may lack accuracy or suitability when generalised to other settings, thereby introducing further risks. The WHO report highlights that even research utilising anonymised or publicly available data can give rise to ethical concerns. AI systems possess the capability to discern patterns and generate predictions from datasets in manners that were unforeseen at the time of the original data collection.

WHO categorises AI-related health research into three distinct categories:

• Health research employing artificial intelligence involves the analysis of extensive datasets to uncover patterns associated with diseases.
• Research utilising AI tools and technologies can enhance various aspects of the research process. AI may assist in developing hypotheses, identifying participants, or producing research outputs.
• Research on AI tools and technologies examines the safety, effectiveness, utility, or implementation of AI systems utilised in healthcare.
• The report says each category can raise different ethical issues, so a single approach to oversight may not be sufficient.

Research ethics committees, or RECs, play a crucial role in safeguarding the interests of research participants. However, the World Health Organization indicates that these committees may require enhanced expertise, training, and resources to effectively evaluate the growing complexity of AI studies. Certain AI research might not be subject to conventional ethics review, as it may not directly engage human participants or may utilise anonymised data. However, the lack of direct interaction does not inherently imply the absence of ethical risks. WHO therefore indicates that oversight may need to encompass a broader range of entities beyond RECs, with researchers, funders, data access committees, scientific journals, professional societies, and regulatory agencies also contributing to this effort.

WHO additionally emphasises the necessity for:

• Enhanced ethics training and certification for researchers
• Early identification of ethical risks
• Enhanced clarity regarding identified risks
• Responsible treatment of data workers
• Enhanced communication and increased public engagement

WHO has also emphasised concerns regarding AI health research in low and middle-income countries, where a significant portion of technology development remains concentrated in higher-income settings. The report indicates that this can pose risks when research is undertaken involving local populations or their data without adequate local participation, expertise, or leadership. Enhancing research capacity and promoting equitable participation will be crucial as the utilisation of AI in health research expands. India already possesses a framework in this domain.

The Indian Council of Medical Research published ethical guidelines for AI in biomedical research and healthcare in 2023, addressing critical issues such as autonomy, safety, data privacy, accountability, fairness, equity, and informed consent. WHO characterises its latest report as a foundational document, acknowledging the swift evolution of AI technologies and their applications. The organization asserts that forthcoming standards and governance frameworks must adapt in tandem with advancements, ensuring that accelerated health innovation does not compromise human rights, equity, and public trust.