AI for Affordable Healthcare

Research papers

Research papers are invited which address challenges important to the real-world deployment of AI in healthcare, such as:

• Limited data, particularly for rare pathologies or for underrepresented populations
• Limited annotation and variability in annotations
• Combining data from different modalities, including handling missing data
• Domain transfer between institutions, systems/technologies, languages and populations
• Interpretability and transparency of solutions
• Secure AI – detecting when patient-sensitive information has been learned by the model
• Safe AI - detecting failure cases with confidence
• Bringing together knowledge bases and data-driven techniques
• Treatment of bias in historical data, and separation of correlation versus causality
• Continual learning and knowledge sharing between institutions
• AI for developing countries – building robust and affordable solutions which can be used with limited technology and healthcare professional support

White papers

We are also welcoming white papers which discuss possible solutions or methodologies for challenges in healthcare. Proposals may:

• Address known challenges or identify new challenges
• Draw on past or ongoing research / real-world experience
• Describe how ideas successfully applied in other domains might be transferable to healthcare

We encourage co-authorship between AI researchers and medical experts. The challenge should be clearly described and the (AI) solution outlined at a system level, with consideration of the economic and patient outcome benefits. Submissions addressing challenges relevant to Ethiopia will be of particular interest (see resources such as the World Health Organisation page on Ethiopia).

Guidelines

Submissions should follow the ICLR 2020 format using Latex style files provided at:

https://github.com/ICLR/Master-Template/blob/master/archive/iclr2020.zip

Submissions will be limited to 5 pages of content and 1 page of references. Papers can contain appendices after the sixth page but reviewers will not be obligated to read these.

We encourage papers with no more than 4 pages where this is required for the purpose of dual submission.

Submission

Submit your paper on OpenReview by February 9th, 2020 at 11:59PM Pacific Time.

https://openreview.net/group?id=ICLR.cc/2020/Workshop/AI4AH

All submissions will undergo double-blind review. Papers should be anonymised, including removing names, affiliations, contact information, and ensuring that references to your own previous work are in the third person.

Accepted Papers

Authors of accepted papers will be notified on February 25th, 2020. Upon notification, we ask that authors of accepted papers address review comments, host their paper on the web (e.g. on arXiv), and then submit a link to where the paper is hosted to the workshop organizers by March 19th, 2020. This website will then host a collection of these links. This does not constitute an archival publication and these papers can be published elsewhere.

All accepted papers will be presented as posters, and selected papers will be presented as oral talks.