United NationsDepartment of Economic and Social Affairs Sustainable Development

ITU-WHO Focus Group on Artificial Intelligence for Health (FG-AI4H)

    Description
    Intro

    The ITU/WHO Focus Group on “AI for Health” (FG-AI4H) was established in July 2018 to develop international evaluation standards for AI solutions in health. The Focus Group works at the interface of multiple fields (e.g., machine learning/AI, medicine, regulation, public health, statistics, and ethics) and includes other decision-makers who value a standardized and transparent benchmarking framework. The international approach offers the opportunity to concentrate diverse national expertise in standardization frameworks on a global level.

    Description

    The overall objectives of the FG-AI4H are to tap this network of international expertise to create (a) guidelines (i.e., document-form “deliverables”) for the evaluation of AI for health and (b) to create an online open source platform and complementary tools for the benchmarking of AI for health, including packages for data acquisition, data storage, annotation, prediction evaluation, reporting. A main output of FG-AI4H are guideline documents hereafter known as “deliverables” (Figure under "other sources"). The deliverables provide the requirements needed to establish a benchmarking process of AI for health. Thematically the work is organized in Topics Groups: Cardiovascular disease management (Sub-topic: Risk Predictions) Dermatology Dental diagnostics and digital dentistry ​Primary and secondary diabetes prediction AI for Endoscopy AI-based detection of falsified medicine Falls among the elderly Histopathology Malaria detection Outbreak detection Ophthalmology Psychiatry Radiology Snakebite and snake identification Symptom assessment Volumetric chest computed tomography Diagnoses of bacterial infection and anti-microbial resistance (AMR) Maternal and child health

    Contribution to SDG Implementation

    Adopted in 2015 by all United Nations Member States, the Sustainable Development Goals (SDG) are an urgent call to action for shared peace and prosperity through improving health, education, reducing inequality and economic growth. The third SDG is dedicated to “Good Health and Well-Being.”. Additionally, the work of FG-AI4H will help achieve gender equality and reduce inequalities by ensuring that AI models work well for all as well as making access to expertise much more accessible.

    Implementation methodologies

    The deliverables represent a collective effort made by members of FG-AI4H. As the collaboration is an ongoing procedure, iterative versions of the deliverables are presented at each bi-monthly meeting. Each topic group and working group produces multiple deliverables. The former include topic description documents; the latter include guidelines on ethical considerations, regulatory considerations (best practices specification), requirements specifications, software lifecycle specifications, data specifications, AI training best practices specifications, evaluation specifications, scale-up/adoption, and FG-AI4H applications and platforms. Deliverables categories 00 Overview of the FG-AI4H deliverables 01 AI4H ethics considerations 02 AI4H regulatory best practices 03 AI4H requirements specification 04 AI software life cycle specification 05 Data specification 06 AI training best practices specification 07 AI4H evaluation considerations 08 AI4H scale-up and adoption 09 AI4H applications and platforms 10 AI4H use cases: Topic description documents

    Results

    The deliverables represent a collective effort made by members of FG-AI4H. As the collaboration is an ongoing procedure, iterative versions of the deliverables are presented at each bi-monthly meeting. Each topic group and working group produces multiple deliverables. The former include topic description documents; the latter include guidelines on ethical considerations, regulatory considerations (best practices specification), requirements specifications, software lifecycle specifications, data specifications, AI training best practices specifications, evaluation specifications, scale-up/adoption, and FG-AI4H applications and platforms. Deliverables categories 00 Overview of the FG-AI4H deliverables 01 AI4H ethics considerations 02 AI4H regulatory best practices 03 AI4H requirements specification 04 AI software life cycle specification 05 Data specification 06 AI training best practices specification 07 AI4H evaluation considerations 08 AI4H scale-up and adoption 09 AI4H applications and platforms 10 AI4H use cases: Topic description documents

    Factors and Constraints

    Key conditions that help lead to the work and progress of the Focus Group include the participation of national and regional medical device regulators, public health agencies, medical professionals, AI developers, ITU-WHO inter-agency collaboration and many others.

    Sustainability and replicability

    The overall objectives of the FG-AI4H are to tap this network of international expertise to create (a) guidelines (i.e., document-form “deliverables”) for the evaluation of AI for health and (b) to create an online platform and complementary tools for the benchmarking of AI for health. As such the output are explicitly designed for replicability.

    COVID-19 Impact

    FG-AI4H Ad-hoc Group on “Digital Technologies for COVID Health Emergencies” (AHG-DT4HE) reviews the role of AI (and other digital technologies) in combating COVID-19 throughout an epidemic’s life cycle.

    Contact Name
    Simão
    False
    Share
    Organization/entity
    International Telecommunication Union and World Health Organization
    SDGs
    Geographical coverage

    The ITU/WHO Focus Group on “AI for Health” (FG-AI4H) was established to develop international evaluation standards for AI solutions in health. The international approach allows diverse national expertise in standardization frameworks on a global level

    Timeline
    25 September 2018 (start date)
    24 September 2021 (date of completion)
    Countries
    Switzerland
    Switzerland
    Partnership
    N/A
    Contact Information

    Simão, ITU-WHO Focus Group on Artificial Intelligence for Health