WHO releases key regulatory considerations on AI for health

The considerations will help to develop new or existing guidance on AI

The World Health Organization (WHO) has published new key regulatory considerations on artificial intelligence (AI) for health to emphasise the importance of establishing the safety and effectiveness of AI systems.

The new considerations aim to help governments and regulatory authorities to develop new guidance or adapt existing guidance on AI at national or regional levels.

With AI rapidly being deployed, WHO has recognised the benefits it has had for health outcomes by strengthening clinical trials, improving medical diagnosis, treatment, self-care and person-centred care, as well as supporting health care professionals’ (HCPs) knowledge, skills and competencies.

Despite this, some AI technologies are being deployed without a full understanding of how they perform, which could potentially harm end-users, including HCPs and patients.

AI health technologies have access to sensitive personal information that can come from clinical settings and user interactions, which requires legal and regulatory frameworks for safeguarding privacy, security and integrity.

In an effort to responsibly manage the rapid rise of AI health technologies, WHO has outlined six areas of regulation for AI for health, including the importance of transparency and documentation, risk management, validating data, intended use, data quality, privacy and data protection, and fostering collaboration.

With better regulation, managing the risks of AI amplifying biases, inaccuracies or failures in training data, including accurate representation of diversity in populations, can be improved in complex AI systems.

Regulations can help to ensure attributes, including gender, race and ethnicity, of people featured in training data are reported and intentionally representative.

Dr Tedros Ghebreyesus, WHO director-general, said: “AI holds great promise for health, but also comes with serious challenges, including unethical data collection, cybersecurity threats and amplifying biases or misinformation.

“This new guidance will support countries to regulate AI effectively to harness its potential, whether in treating cancer or detecting tuberculosis, while minimising the risks.”