Initiative involves joint work with India, Russia and South Africa and proposes the use of artificial intelligence resources to control the pandemic
A Unicamp project, developed in partnership with Unesp, USP and Insper and also with universities in India, Russia and South Africa, was one of the 12 selected by the BRICS Covid-19 Public Call (CNPq/MCTI/BRICS-STI nº 19 /2020), carried out by the Ministry of Science, Technology and Innovations (MCTI), through CNPq. The initiative aims to develop artificial intelligence tools in the country that can read different data regarding Covid-19, such as information obtained from medical records, laboratory and imaging tests, geographic information, data on people's movements. , among others, to expand and speed up the capacity for diagnosis and calculation of the incidence of the disease. The project will receive an investment of R$700 in resources allocated by MCTI and the National Health Funds (FNS) and National Scientific and Technological Development Funds.
The project is coordinated in the country by Professor Esther Colombini, from the Computing Institute (IC) at Unicamp. From the University, professors Maurício Perroud Junior and Andrei Sposito, from the Faculty of Medical Sciences (FCM), also join the team. From Unesp, professors Alexandre Simões, from the Sorocaba Institute of Science and Technology (ICTS) and Carlos Magno Fortaleza, from the Faculty of Medicine of Botucatu (FMB), participate. From USP, Fátima Nunes Marques, professor at the School of Arts, Sciences and Humanities (USP Leste) and João Eduardo Ferreira, professor at the Institute of Mathematics and Statistics (IME), are working. In addition to researchers from the three state universities in São Paulo, the group also includes professor Danilo Carlotti, from Insper.
According to Esther Colombini, the advantage of forming this partnership between several institutions, spread across the State of São Paulo, is the possibility of involving several data sources in the project from which relevant information can be extracted to determine the prevalence of Covid-19. This is the case, for example, of data that can be obtained from the State's university hospitals. "Our proposal is to use multimodal data, both medical images and patient symptomatology data, available in medical records, using natural language processing, and various exams. We have several indications that, with some common blood tests, associated with others aspects and geographic location, by crossing information with infection rates in the regions, we can increase the accuracy of the models to detect whether people are infected or not, and also whether they have a milder or more severe infection", explains the coordinator of the project.
In this way, the initiative can contribute not only to faster and more accurate diagnoses of the disease, but also to find new relationships between symptoms, risk factors and the environment, which can favor the diagnosis of other diseases. "Data mining and the use of artificial intelligence are essential for detecting early signs of emerging diseases, allowing coordinated action by countries in accordance with the International Health Regulations. Initiatives like this can control infections before they become pandemics", analyzes Carlos Magno Fortaleza, professor at FMB/Unesp.
Another objective of the project is that it offers researchers in different areas of knowledge and public authorities instruments that improve the decision-making processes of public managers and also contribute to the development of efficient policies to combat the pandemic. "The idea is to integrate artificial intelligence at the tip of the problem, which can identify the spread of the disease, with public policy decisions that are taken at the last level and that impact us in everything", comments Esther Colombini.
International cooperation and new research partnerships
The researchers' initiative does not end with the development of this system that uses artificial intelligence to analyze data regarding the coronavirus. It includes developments related to controlling the pandemic and evaluating public policies that will be implemented by other BRICS countries, in this case India, South Africa and Russia.
Also based on artificial intelligence resources, researchers in India intend to combine data on people's movement, obtained through the use of applications on cell phones, with health information, to map outbreaks of coronavirus spread and point out locations that need greater attention. South Africans will focus on this mapping of areas with the greatest health risk to build models and logistical systems for distributing medical supplies, health professionals and vaccines, so that collapses in health systems are avoided. In Russia, researchers will be able to analyze the quality of public policies adopted to contain the virus, checking in which scenarios lockdowns and openings were effective and how good decisions can be replicated in other locations.
"It is important to remember that we are talking about countries like Brazil, India and Russia, which are three countries with continental dimensions, with very different characteristics. It is a very important issue for the project that these countries have continental dimensions. At the moment When we developed the project, they were the countries most affected by the coronavirus, apart from the United States. And South Africa, in particular, has challenges in logistical terms", details Esther Colombini.
The professor also clarifies that the main objective of this cooperation is that the results obtained by each country in their specific projects serve as input for the other members, at the same time that the expertise of each one is shared with the others. In the end, each country contributes by building part of a large project that is fully developed in the four countries. She also emphasizes the importance of establishing new research networks for future initiatives: "This arrangement with these institutions was set up for this project, but this is interesting because it brings collaboration with institutions that we don't necessarily already cooperate with."
