Artificial Intelligence against COVID-19

ICA Laboratory at PUC-Rio and Cenpes / Petrobras

This page is an initiative of the Laboratory of Applied Computational Intelligence (ICA) and Cenpes / Petrobras, partners for 20 years in the research and development of artificial intelligence projects for oil and gas sector.
The artificial intelligence research against coronavirus presented on this website results from advanced studies in underwater image recognition and natural language processing on technical / scientific bases, proposed by researchers from Cenpes Reservoir Engineering Management and developed with ICA under the Cooperation Term called BIG OIL – Data Science for Oil & Gas.

In March 2020, the World Health Organization (WHO) declared the new disease caused by the coronavirus, COVID-19, as a pandemic. The consequences of this pandemic go beyond the loss of human life, extending to profound impacts on the world economy. In this context, the Laboratory of Applied Computational Intelligence (ICA), in partnership with Cenpes / Petrobras and with the support of several specialists from health area seeks to identify ways to characterize and detect coronavirus occurrences, applying advanced models of Artificial Intelligence (AI). This page presents and displays results of this research in an interactive way, as well as invites scientists and doctors to collaborate, either through suggestions, relevant data to the research and, mainly, requests for AI applications that are related to our work. First, one of the project goals are to gather all scientific information that contains references to the new virus. Such information includes biomolecular aspects and development of antiviral treatments. For this purpose, we applied the recent advances in Natural Language Processing (NLP) and other AI techniques to generate new insights in a collection of academic literature on COVID-19, SARS-CoV-2 and related coronaviruses. As such, we hope to obtain answers to questions and connect information about this content in support of COVID-19’s ongoing response efforts around the world. Preliminarily, an unsupervised word / doc2vec style analysis of the COVID publication data is made available. We believe that the knowledge for future discoveries can be supported by these previous publications. Another goal of the project is to use X-ray images and personal information from several patients to identify the presence of pulmonary complications caused by the virus in simple X-ray images. The project uses Deep Learning Neural Networks models to the contaminated information over time, such as: sex, age, pre-existing diseases, location (country and state), clinical conditions, X-ray images and information about the progress of the illness. With all these records, complex neural networks are trained, among other things, to identify patients infected or not with the COVID-19 virus. As an auxiliary form of research, we are providing a channel for sending information on patients admitted to hospitals.