The study on the predictive accuracy of artificial intelligence (AI) Lunit INSIGHT CXR Version 3.0 for pneumonia diagnosis in COVID-19 patients

Abstract

Background: Millions of people in Thailand have been infected and died from the infection of the COVID-19. As a result, the country’s public health system is greatly affected due to the limitation of the number of physicians. Artificial intelligence (AI) is, therefore, used to reduce the working load of physicians in the diagnosis of COVID-19 patients. Objective: To study on the predictive accuracy of AI Lunit INSIGHT CXR Version 3.0 for pneumonia diagnosis in COVID-19 patients. Materials and Methods: This study was a retrospective study. The data was collected from 256 confirmed cases of COVID-19 infection admitted as new patients in the Nimibutr Pre-Admission Centre of the Institute of Neurology, the Ministry of Public Health. They were randomly selected from the database. Seven radiologists and Lunit INSIGHT CXR Version 3.0 software interpret the CXR film to diagnose pneumonia in COVID-19 patients from chest radiographs (CXR). Results: The research results of the diagnosis of pneumonia in patients infected with COVID-19 between from radiologists and using AI Lunit INSIGHT CXR Version 3.0 software revealed 97.87% (95%CI 88.71-99.95%) of sensitivity, 99.04% (95%CI 96.59-99.88%) of specificity, accuracy = 98.83%, positive predictive value (PPV) = 95.83%, and negative predictive value (NPV) = 99.52%, positive likelihood ratio (+LR) = 102.28, negative likelihood ratio (-LR) = 0.02. Conclusion: The artificial intelligence software Lunit INSIGHT CXR Version 3.0 can be used to interpret the diagnosis of pneumonia in patients infected with COVID-19 in order to reduce radiologists’ workloads during the COVID pandemic when medical staff were limited.

Ähnliche Arbeiten