The World Health Organization (WHO) has recently published a document focusing on the essential regulatory considerations for the application of artificial intelligence (AI) in healthcare. This publication emphasizes the need for safety, efficacy, and accessibility of AI systems in healthcare, and encourages collaboration among various stakeholders, including developers, regulators, manufacturers, healthcare professionals, and patients.
Key aspects of the publication include:
AI’s Potential in Healthcare: AI, with its advanced analytic techniques like machine learning, logic-based approaches, and statistical methods, has the potential to revolutionize healthcare. It can enhance clinical trials, medical diagnosis and treatment, self-care, person-centred care, and augment the skills of healthcare professionals.
Regulatory Challenges and Needs: The deployment of AI technologies in healthcare is happening rapidly, and there is a need for comprehensive understanding of their potential impacts. Concerns about the use of health data by AI systems highlight the need for strong legal and regulatory frameworks to ensure privacy, security, and data integrity.
WHO’s Stance on AI in Healthcare: Dr. Tedros Adhanom Ghebreyesus, WHO Director-General, acknowledges AI’s promise in healthcare but also points out challenges like unethical data collection, cybersecurity threats, and the risk of amplifying biases or misinformation. The guidance aims to help countries regulate AI effectively to harness its potential while minimizing risks.
Key Areas for Regulating AI in Health: The publication outlines six key areas for regulating AI in health, including transparency and documentation, risk management, external validation, data quality, understanding jurisdictional scope and consent requirements, and encouraging collaboration among stakeholders.
Addressing Biases and Inaccuracies: The document stresses the importance of regulatory measures to address biases and inaccuracies in AI systems. This includes mandating the reporting of demographic attributes in training data to ensure datasets are representative of diverse populations.
Guidelines for Governments and Regulatory Authorities: The WHO publication provides fundamental principles for governments and regulatory authorities to develop new guidelines or adapt existing ones related to AI at national and regional levels.