Health 1, 2018

Policy on the Use of AI in the Healthcare Sector

United Arab Emirates

Actors

Abu Dhabi Department of Health

Tags

Transparency, safety, privacy, accountability,

Resources


The policy was issued by the Department of Health (DOH), which is the regulatory body of the health system in the Emirate of Abu Dhabi. Recognising the central role AI can play across the healthcare sector and seeking to capitalise on the benefits to be gained from its use the DOH supports and encourages the development, use and adoption of customized AI technologies and software for healthcare in Abu Dhabi, while seeking to minimize its risks.
The policy opens with a definition of key terms, including that of ‘graceful degradation’, which plays an important role in the policy and is understood as “the ability of a computer, machine, electronic system or network to maintain limited functionality even when a large portion of it has been destroyed or rendered inoperative.”
The purpose of the policy is to outline the Emirates’ vision for AI in healthcare and to outline key roles and responsibilities of stakeholders. The policy applies to licensed healthcare providers, pharmaceutical manufacturers, health insurers, healthcare researchers, as well as anyone who uses Abu Dhabi-based population or patient clinical or non-clinical data in an AI endeavour.
The policy sets out 6 key principles:
[ul]
  • Transparency: verifiable and explainable AI to ascertain the reasons of any system failures.

  • User Assistance / Supportive technology: AI should assist users in making intelligent decisions.

  • Safety and Security: Robust and dependable AI is essential to ensure it doesn’t harm the lives or bodies of users or third parties.

  • Privacy: AIs should not infringe the privacy rights of users or third parties.

  • Ethics: human dignity and individual autonomy should be respected in AI R&D.

  • Accountability: researchers and developers need to be accountable to users and other stakeholders, disclosing relevant information and maintaining adequate communications with stakeholders.

  • [/ul]
    DOH is committed to maximising the benefits and minimising the risks of AI. In order to achieve this, it will create a regulatory framework building on its 6 key principles. 
    The policy also formulates a set minimum acceptable requirements against healthcare AI applications. These include the requirement of AI systems to be robust and responsive evidenced by certification from international agencies, compliance with Abu Dhabi Smart Solutions & Services Authority regulations, and auditable validation statements. AI tools must also have mechanisms for graceful degradation built-in, and should be continuously updated and improved on the basis of user feedback. Systems should also be subjected to audits but the criteria for such audits are not outlined. AI tools must also comply with Health Information Exchange Policy (HIEP) regulations, which includes privacy and transparency requirements. However, the HIEP has only one provision on transparency, establishing it as a guiding principle and calling for broad stakeholder engagement. 
    The policy sets out requirements for non-patient users/providers of AI services as well, including the requirement of guidelines about access to and use of patient information, adequate education on AI tools, reporting of audits to DOH, compliance with all regulatory requirements, and the submission of end-user feedback to DOH. 
    A monitoring and evaluation framework on the effectiveness of AI in healthcare is to be developed, focusing on inputs, activities, outputs and outcomes.

    AI Governance
    Database

    This database is an information resource about global governance activities related to artificial intelligence. It is designed as a tool to help researchers, innovators, and policymakers better understand how AI governance is developing around the world.

    We're collecting more examples as we find them. If you'd like to add yours to this map, please suggest an entry.

    Suggest an entry
    • Filter by:
    • Current filters:
    256 entries are listed below
    • Sort by: