ARTIFICIAL INTELLIGENCE IN DIAGNOSING COVID-19 PNEUMONIA AND PULMONARY TUBERCULOSIS IN THE KYRGYZ REPUBLIC

Abstract

Nowadays, the necessity to control lung diseases such as COVID-19 caused by the SARS-CoV-2 virus and tuberculosis is obvious. One of the most important areas of this work is rapid and accurate diagnostics, including lung imaging based on artificial intelligence (AI). Objective. The aim of the paper is to test AI for detecting COVID-19 pneumonia and pulmonary tuberculosis based on digital X-ray patterns. Materials and Methods. The study included several stages. 1. Development of an AI model for detecting COVID-19 pneumonia and pulmonary tuberculosis. 2. Creation of a test X-ray data base. 3. Interpretation of data by radiologists. 4. Use of AI for diagnosing COVID-19 pneumonia and pulmonary tuberculosis. Results. AI demonstrated good prognostic ability (sensitivity – 88.31 % and 83.33 %, specificity – 96.67 % and 97.78 % for pneumonia and pulmonary tuberculosis, respectively). AI effectively processes and analyzes big data, which saves doctors’ time. However, in order to ensure greater safety, healthcare professionals should bear responsibility for the final diagnosis. The collaboration between radiologists and AI seems to be desirable. AI can be an auxiliary tool in conditions of high workload or shortage of specialists, as it can improve the accuracy of radiological reports and ensure their timeliness.

Ähnliche Arbeiten