FG-AI4H Working Group & Ad-hoc Group ToRs

Ch​air:

Pat Baird (Philips)

​Vice-​​chair:

Luis Oala (Fraunhofer HHI, DE)​

ToR 

Scope

This WG is concerned with the identification and definition of methods for data and AI solution assessment, while the implementation of said methods is coordinated with the WG Data and AI Solution Handling.

The modus operandi for this WG is in close cooperation with the TGs, in particular for the definition of task specific assessment methods for data and the AI solutions, as well as the other WGs that work on the technical implementation of the standardized assessment framework.

Specific issues to be addressed include

The WG pursues the following tasks:

a) Identify r​​elevant categories of AI solution assessment; e.g. performance, stability, or transparency among others; in close cooperation with the topic drivers from the TGs.

b) Specify concrete methods for AI solution assessment based on (1) task specific considerations and (2) state-of-the art machine learning research. This collection of methods shall include comments on benefits and limits of the included methods.

c) Identify relevant categories of data assessment; e.g. balance, representation or measurement quality among others; in close cooperation with the topic drivers from the TGs.

d) Specify concrete methods for data assessment based on (1) task specific considerations and (2) state-of-the art machine learning research. This collection of methods shall include comments on benefits and limits of the included methods.

Deliv​erabl​es

a) Recommendations for categories of data and AI solution assessment ​

b) Recommendations for data and AI solution assessment methods with comments on benefits and limits of the included methods 

c) Implementation specifications of B. to WG Data and AI Solution Handling​

d) Progress reports to FG 

Relationships

• ​T​opic drivers (members of FG-AI4H) for the different topic areas​
  
• ​Chairs of the other working groups of FG-AI4H
• ITU secretariat
• WHO representatives
• Workshop hosts (institutions, academy, industry)
  

Chair

Andreas Reis (WHO)

​Motiv​​ation

Digital technologies, machine learning and Artificial Intelligence (AI) are revolutionizing the fields ​of medicine, research and public health in an unprecedented manner. While holding great promise, this rapidly developing field raises a number of ethical, legal, and social concerns. There is an urgent need to develop harmonised ethics guidance for the design and implementation of AI in global health. Moreover, to secure AI benefits at the global scale, a new collaborative research agenda should be established. WG-Ethics will address these ethical concerns and provide guidance to the users of AI.

Sc​​ope

Producing harmonised ethics guidance for the design and implementation of AI in global health; addressing the ethical development and use of these technologies, including how low- and middle-income countries will benefit from AI developments.

Specific questio​​ns to be addressed include:

• Equitable access
• Privacy
• ​Appropriate uses and users
• Liability
• Bias
• Inclusiveness 
  

Deliverab​​les

• Develop key documentation that synthesizes discussions conducted among WG-Ethics members and provides ethical guidance for the users of AI for health.
• ​Providing input into WHO guideline development processes.
• Sharing of drafts that result from the work of the WHO Experts Group on Ethics.
• Providing guidance for ethical considerations of AI for health topics.
  

Relation​​​ships

• Chairs and experts of the other working groups of FG-AI4H
• ITU secretariat and WHO representatives
• WG members with relevant expertise
• Workshops to include participants from government agencies, academic institutions, NGOs and national ethics committees
  

Cha​​ir

Naomi Lee (The Lancet, UK)

Vice-c​​h​​​airs:​

• Michael Berensmann (Federal Institute for Drugs and Medical Devices, Germany)
  
• Robin Seidel (Federal Institute for Drugs and Medical Devices, Germany)​
  
• Liang Hong (Center for Medical Device Evaluation, National Medical Products Administration​, C​​hina)​
  
• Chandrashekar Ranga (CDSCO, India)​
  
• Paolo Alcini (European Medicines A​​gency, EU)​, observer role.
  
  

Structure

The diagram below illustrates the proposed structure of the Working Group "Regulatory considerations on AI for health" (WG-RC).

The WG has o​ne chair nominated by the Focus Group. ITU or WHO members can each nominate a vice-chair from their regulatory bodies*. ​Vice-chair nominations are confirmed either at Focus Group meetings or by the FG management in-between meetings.​

* Regulatory bodies are meant here to include national and regional organizations that have in their mandate the responsibility for assessing the safety and efficacy of AI for health technologies within their jurisdictions.

Scope and des​​​cription

The working group "Regulatory considerations for AI" will help guide the FG in navigating the regulatory landscape; facilitating contacts, information exchange, and collaborative opportunities of the FG with regulatory bodies. The working group will help outline key regulatory considerations that are relevant to regulatory agencies for AI development. The working group will also help define ways to successfully benchmark AI for health algorithms.

​Specific issues to be addressed include:

• How can the FG work together with regulatory bodies to ensure that the procedures, benchmarks, and validation principles are useful and relevant for regulators and other stakeholders? This will further facilitate a wider adoption of AI in health.
• Tasks include, but are not limited to:
  
  • Organize networking opportunities to facilitate information and knowledge exchange between experts, regulators, and different stakeholders
  • Identify regulatory challenges and opportunities
  • Develop an outline of shared key considerations to consider for the regulation of AI in health

​Deliverabl​e​​s and timeframe

• ​In close collaboration with the WHO, the working group will develop appropriate key regulatory considerations document that will serve as consensus elements to establish policies and regulations that ensure the safe and appropriate development and adoption and use of AI in health.
  (Expected timeframe: initial draft by the summer of 2019. Finalize by the end of 2019.)
  
• Outline of opportunities and challenges for the development and adoption of AI in health and will propose demonstration projects and workshops to help further the understanding of all stakeholders.
  (Expected timeframe: By the end of 2019 for documentation. TBD for demonstration projects.)
  

Representations and​ interactions

• Chairs and experts of the other working groups of FG-AI4H
• ITU secretariat and WHO representatives
• Invite WG members from different regulatory agencies with relevant expertise.
• Workshops to include participants from different stakeholders (academic institutions, industry (biotech, pharma, informatics, etc.), regulatory agencies, bioethics, patient advocacy groups and other stakeholders).​​
  

Co-chairs

Ana R. Cinnamond (PAHO/WHO), Shan Xu (CAICT, China)

​Scope and description 

Initially, the ad-hoc group (AHG) will collect effective ways and cases on AI and other digital technologies to combat COVID-19 covering the entire cycle of an epidemic emergency, encompassing the following: prevention and preparedness, outbreak early detection, surveillance and response, recovery, rehabilitation, mitigation, etc (Reference: Emergency Risk Management/ Disaster Risk Reduction of the ERF Sendai Framework​​). The outputs are expected to evolve towards a more generalizable mechanism on the health emergency continuum, eventually applicable to other pandemics.

Specific questio​​ns to be addressed include:

• COVID-19
• Health emergency cycle
• Digital technologies including AI on epidemic emergencies
  

Deliverab​​les

• Nurture the creation of a network of experts that can be a reference group for addressing the use of AI and other digital technologies for tackling health emergencies such as the COVID-19 pandemic.
• ​Develop a collection of best practice on the use of AI and other digital technologies and best practices that have been used during each of the stages of the COVID-19 emergency cycle and that can be referenced globally and build suitable mechanisms for sharing these experiences.
• Develop collaborations with relevant FG-AI4H working groups and topic groups (e.g. WG-Ethics, TG-Outbreak, TG-Symptoms, TG-DiagnosticCT, etc.).
• Provide, as appropriate and upon request, input to PAHO development processes for guidelines on the use of AI and other digital technologies for tacking health emergencies.
  

Representations and interactions

• Chairs and experts of the FG-AI4H
• WG and TG members with relevant expertise and interest
• ITU and WHO experts
  
• G20 and other international mechanisms​
• Workshops to include participants from government agencies, academic institutions, NGOs and national healthcommittees
  

Declaration of conflict of interest

Participants in this ad-hoc gr​oup will follow the procedure in FG-AI4H F-105 (Zanzibar, September 2019), and ToRs for the WG-Experts and call for experts and Annex A, Application form; Conflict of interest form​.