Open AccessArtificial Intelligence in Healthcare: Validating an AI Algorithm in Health Institutions in the COVID-19 Pandemic (a Use Case)
Author(s) -
Ivan A. Blokhin,
Sergey P. Morozov,
Valeria Chernina,
Anna Andreychenko,
I.V. Shakhabov,
A.V. Smyshlyaev,
Victor А. Gombolevskiy
Publication year - 2021
Publication title -
monitoring obŝestvennogo mneniâ: èkonomičeskie i socialʹnye peremeny
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.359
H-Index - 8
eISSN - 2219-5467
pISSN - 1815-8617
DOI - 10.14515/monitoring.2021.1.1736
Subject(s) - pandemic , covid-19 , health care , artificial intelligence , predictive value , computer science , public health , sample (material) , medicine , machine learning , medical emergency , algorithm , nursing , pathology , political science , chemistry , disease , chromatography , infectious disease (medical specialty) , law
The paper considers new challenges related to public health. Action is needed to improve access to healthcare while maintaining its quality. The introduction of AI-based automated data analysis systems can be a solution to that. The present study seeks to assess the use of AI in outpatient care to detect pathological changes in the lungs typical of a coronavirus amidst the pandemic. The sample size was 600 patients. The results were statistically and analytically processed. The sensitivity attained 94%; the specificity, accuracy and the area under the ROC curve were 77%, 83%, and 87%, respectively. The negative predictive value was 97%; the positive predictive value was 66%. The data obtained show that the algorithm separates the CT scan results having no abnormalities in the lungs. The authors conclude that the usage of AI technologies helped to improve diagnostic accuracy during the COVID-19 pandemic. Artificial intelligence algorithms can also work with patients in non-pandemic times, thus improving healthcare access.