All together in an international virtual challenge to fight Covid-19 thanks to the "Covid CXR" Hackathon project promoted by Italian Institute of Technology, from the Bruno Kessler Foundation and fromUniversity of Modena and Reggio Emilia (UNIMORE).
The call took place on February 1st during the international event "Artificial Intelligence & Cybersecurity for human health" organized by Italy, Israel and the United Arab Emirates which took place at the amphitheater of Italy Pavilion.
The forum falls under the series of events organized by the Italian Pavilion for "Health & Wellness Week" there thematic week of Expo 2020 Dubai, dedicated to health and well-being which took place from January 27 to February 2.
"Accuracy of prognosis" and "explainability". These are the basic elements of the “Covid CXR” Hackathon. The goal is to support healthcare professionals in the prognosis activity starting from chest radiographs and clinical data collected during the hospital triage phase.
Hackathon "Covid CXR" at Expo 2020 Dubai
The health emergency from Covid-19 highlighted the capacity of technological innovations in the health field and the need to support international research in new diagnostic methods and pharmaceutical therapies.
In this scenario, data is at the center of research and therefore international cooperation in sharing data in full respect of privacy is very important.
The project confirms the importance of sharing data "Covid CXR" hackathon presented to Expo 2020 Dubai. It was an opportunity to call students, doctoral students and teams of researchers to participate in an international virtual challenge to fight Covid-19.
Hackathon "Covid CXR" was born in 2020 during the first wave of Covid-19 in Northern Italy thanks to the collaboration of engineers and doctors of various Italian hospitals coordinated by Italian Diagnostic Center.
"We have achieved two important results: a database created by the CDI that contains various information on patients shared with the scientific community and the development of tools that do things that are more or less similar but with different approaches". He said in his speech Diego Sona, researcher of the Data Science for Health unit (FBK Digital Health & Wellbeing Center).
“This challenge aims to demonstrate - he added - that AI is almost ready to be used and applied to real health problems. Clearly this will raise ethical issues and one way to address these is to make AI tools transparent through the concept of explainability of the model that the doctor can use to make an AI-supported decision, for this we have introduced a reward for the best methodology that explains the predictions of AI. "
How to participate in the "Covid CXR" Hackathon
L'hackathon it will last a month and is open to students, doctoral students and teams of researchers who will have to develop systems capable of processing real data, concerning the first hospitalization of COVID patients and coming from various healthcare facilities in Northern Italy active during the first outbreaks.
Therefore, these are images whose quality and format are very variable and whose clinical data may be incomplete.
The solutions with the best "accuracy of prognosis" and the best "explainability" will be rewarded.
Participants will have to find and present solutions with better accuracy in prognosis, but also enriched with characteristics of explainability and transparency to be acceptable and usable by health experts, who are not necessarily also artificial intelligence experts.
The contributions will be evaluated by a group of doctors and computer scientists.
"Covid CXR" hackathon wants to address in an innovative way, through AI, one of the main issues raised during the Covid-19 emergency, namely the additional care burden for health infrastructures, which often approaches or exceeds treatment capacities.
Automated or semi-automated techniques developed using machine learning could help clinicians distinguish between patients who can be treated safely at home, and those who may require intensive care in the hospital, improving planning and more effective allocation. of available resources.
"The Covid CXR Hackathon was designed to transform the pandemic experience into a stimulus for the scientific community to demonstrate that the development of automatic intelligent systems, both for image recognition and other types of data, is ready for comparison. with the most pressing problems of our society. The comparison with clinicians, companies and industrial realities is fundamental ". He has declared Alessio Del Bue, Head of the PAVIS research line (Pattern Analysis and computer VISion) of the Italian Institute of Technology.
Del Bue during his speech presented the stages and methods of participation in the “Covid CXR” Hackathon challenge.
“We already have 41 teams from around the world who signed up 3 days ago. Thanks to today's event (1 February, ed) we hope for an increase with an ever increasing number of data available. The presentation will start on February 15th and will end on March 1st. Subsequently, on March 3, teams will have to submit a description regarding the "explainability" of how their solution worked. The awarding of prizes will follow after evaluation by the scientific committee made up of medical and clinical informatics ".
As Del Bue recalled at the end of his speech, the realization of this project was possible thanks to the collaboration of Italian centers of excellence.
The “Covid CXR” Hackathon proposal is supported by Labs of Learning and Intelligence Systems and units ELLIS of Genoa (Italian Institute of Technology and University of Genoa), the University of Modena and Reggio Emilia and Technion University as well as the Bruno Kessler Foundation.
The Hackathon is sponsored by Bracco Group, from Italian Diagnostic Center (CDI) and from CINI-AIIS ITALIANO (National Laboratory of Artificial Intelligence and Intelligence Systems). The event is also supported by NVIDIA which will provide the computational structures during the hackathon.