One of the major issues raised during the Covid-19 crisis is the additional burden on healthcare infrastructure, often nearing or surpassing treatment capacities.
Automatic or semi-automatic techniques to distinguish patients that can be safely home-treated from those that are likely to require intensive care would allow for better planning and smarter allocation of available resources. Artificial Intelligence, and more specifically machine learning applied on clinical data could be a reliable answer to this demand.
To this end, the Covid CXR Hackathon aims at designing effective solutions based on machine learning and data science supporting the medical doctor to formulate a Covid-19 prognosis from early chest X-ray images and clinical data collected during triage.
One of the main challenges of this hackathon will be the need for processing from several different structures upon first hospitalization. This was done in near-emergency conditions during the first outbreak in Northern Italy. Hence, image quality (and format) is highly variable and clinical data, despite best intentions, is incomplete. Therefore, any developed approach will have to deal with missing data. This hackathon targets finding solutions relying on both sets of data, with a heavy emphasis on image analysis. Solutions not using CXR images will not be considered.
The participants are expected to find and submit solutions with improved accuracy in prognosis but also enriched by explainability and transparency features to be acceptable and usable by health experts, which are not necessarily AI experts too. Therefore, the submissions will be evaluated quantitatively on performance as well as on explainability by a panel of clinician and computer scientists.
The Hackaton will be launched February 1st 2022 at the Dubai Expo 2020 during the Workshop “Artificial Intelligence and Cybersecurity for Health” co-organized by Italy, Israel and UAE.
Hackathon will last one month, check important dates.
The Hackathon proposal is supported by the Labs of Learning and Intelligence Systems, and the ELLIS units of Genoa (Italian Institute of Technology and University of Genova), of University of Modena and Reggio Emilia and Technion University as well as the Bruno Kessler Foundation. The Hackathon is endorsed by Bracco Imaging and Centro Diagnostico Italiano (CDI) and the ITALIAN CINI-AIIS (National Laboratory of Artificial Intelligence and Intelligence Systems). The event is also supported by NVIDIA that will provide Computational Facilities during the hackathon.
Prizes will be granted for the solutions with the best “prognosis accuracy” and best “explainability”.
For additional information on the Hackathon check the following links: