CONF: Autism and artificial intelligence / by Erich Kofmel

The International Telecommunication Union, the specialized agency of the United Nations concerned with information and communication technologies, will be holding a conference on artificial intelligence, "AI for Good Global Summit 2018", from 15-17 May 2018, at ITU in Geneva, Switzerland.

From the description: "The AI for Good series is the leading United Nations platform for dialogue on AI. [...] While the 2017 summit sparked the first ever inclusive global dialogue on beneficial AI, the action-oriented 2018 summit will focus on impactful AI solutions able to yield long-term benefits and help achieve the Sustainable Development Goals. 'Breakthrough teams' will demonstrate [...] new opportunities for AI to help achieve Universal Health Coverage [...]. Teams will propose impactful AI strategies able to be enacted in the near term, guided by an expert audience of mentors representing government, industry, academia and civil society. Strategies will be evaluated by the mentors according to their feasibility and scalability, potential to address truly global challenges, degree of supporting advocacy, and applicability to market failures beyond the scope of government and industry. The exercise will connect AI innovators with public and private-sector decision-makers, building collaboration to take promising strategies forward."

Of particular relevance to autistic persons is that the track on "AI and health" intends to "[i]dentify quick-wins areas and types of AI applications that hold high-potential impact on health outcomes and that are feasible and relatively simple to deploy". Under "Potential domains for AI quick-wins for Public Health", they list a number of workstreams, including "AI for primary care and service delivery", expecting that "Artificial Intelligence can help make certain types of diagnostics almost ubiquitous, sometimes in combination with off-the-shelf cameras or smartphones. Similar tools can help guide primary-care workers through treatments that they might otherwise not have been able [to] offer", among them "[s]ervices for [...] early childhood development disorders e.g., detection of [...] autism", "[p]re-diagnostics, self-assessment and remote screening ([...] mental health, [...] etc.)", and "AI-powered health consultations / telemedicine". Another workstream will discuss chatbots that "can help people with identifying [...] mental conditions".

If you are technologically well versed, please consider registering for this conference to ensure that the autistic perspective with regard to diagnostics and early detection is taken into consideration. Let us know if we can be of assistance:

Otherwise, the entire conference will also be webcast:

Unfortunately, it is not clear if all tracks, such as "AI and health", and workstreams, i.e. "AI for primary care and service delivery", will be webcast simultaneously.

[Update 8 September 2018: As one of the outcomes of the AI and Health track, an ITU Focus Group on Artificial Intelligence for Health (FG-AI4H) was formed, which will meet for the first time on 26-27 September 2018 and is organizing a workshop on "Artifical Intelligence for Health" on 25 September 2018 at the headquarters of the World Health Organization (WHO) in Geneva, Switzerland.

This is the first workshop in a planned series of workshops: "The aim of these workshops is to provide a platform for researchers, engineers, practitioners, entrepreneurs and policy makers to discuss standardization opportunities for the assessment of AI for health solutions and to identify use cases and data required for the evaluation and validation with open benchmarks."

While autism isn't mentioned in the programme of this workshop, and it seems very technical, some issues to be discussed are of concern to autistic people, such as access of AI researchers to databases, including repositories collecting our DNA and genetic and patient information, "for training and testing of the algorithms in the context of evaluation and validation with open benchmarks. It will address the need of open data sources and opportunities for how to make algorithms fair and unbiased, how to promote algorithmic accountability and how to make predictions equitable. [...] Identify who can share, own, control and maintain a variety of datasets and where should this data be stored in order to support 24/7 operational benchmarking infrastructure."

The workshop seems to be open to everyone who is interested, and the registration form includes the option of remote participation. Remote participation is also available for the subsequent FG-AI4H meeting. Places are limited, and applications (both for on-site and remote participation) will be handled on a first-come, first-served basis. Please register as soon as possible.

Furthermore, the FG-AI4H is inviting written contributions for its meeting, "inter alia on (1) the state of the art for artificial intelligence for health (e.g., terms, definitions, concepts, requirements, research gaps, methods, formats, standards ecosystem, applicable assessment / benchmarking frameworks); (2) specific use cases and their standardization; (3) medical AI research and standardization of use of AI in medical devices and diagnostics in healthcare; (4) top health industry issues/‌health emergencies and how AI could address those in a standardized approach; (5) medical data required for testing, training and evaluation of AI4H algorithms, considering various dimensions, e.g. availability, standardization, privacy and ownership."

Please use the template below and send written contributions to the secretariat:

Deadline for registration and written contributions: 19 September 2018

Written contributions will likely also be accepted for future meetings of the FG-AI4H. Any upcoming opportunities and events will be announced on its webpage: ]