Open AccessAn Ontology Based Natural Language Processing Pipeline for Brazilian COVID-19 EMR
Author(s) -
Raquel Gritz,
Rafael Silva Pereira,
Henrique Matheus F. da Silva,
Henrique G. Zatti,
Laura E. A. Viana,
Karol C. S. F. Navarro,
T Dias,
Viviane S. B. Oliveira,
Ricardo Augusto de Souza,
Vinícius Almeida Oliveira,
Manoel Barral Netto,
Fábio Porto
Publication year - 2021
Language(s) - English
Resource type - Conference proceedings
DOI - 10.5753/bresci.2021.15794
Subject(s) - ontology , computer science , pipeline (software) , covid-19 , task (project management) , portuguese , natural language processing , medical record , pandemic , information retrieval , open biomedical ontologies , health records , data science , artificial intelligence , world wide web , upper ontology , disease , health care , linguistics , medicine , infectious disease (medical specialty) , ontology alignment , semantic web , economic growth , pathology , management , epistemology , radiology , programming language , economics , philosophy
COVID-19 became a pandemic infecting more than 100 million people across the world and has been going on for over a year. A huge amount of data has been produced as electronic medical records in the form of textual data because of patient visits. Extracting this information may be very useful in better understanding the COVID-19 disease. However, challenges exist in interpreting the medical records typed as free text as doctors may use different terms to type in their observations. In order to deal with the latter, we created an ontology in Portuguese to describe the terms used in COVID-19 medical records in Brazil. In this paper, we present a brief overview of the ontology and how we are using it as the first step of a more complex NLP task.