Early Detection of Asymptomatic Covid-19 Infection with Artificial Neural Network Model Through Voice Recording of Forced Cough

International Journal on Informatics Visualization: JOIV

article

Aisyah Khairun Nisa¹  I Gede Pasek Suta Wijaya¹  Arik Aranta¹ 

[1]Mataram University

关键词: ANN;    Asymptomatic;    COVID-19;    Forced Cough;    Oversampling;   

DOI  :  10.30630/joiv.7.2.1812

来源: Politeknik Negeri Padang

【 摘 要 】

SARS-CoV-2 is a virus that spreads the infection known as COVID-19, or Coronavirus 2019. According to data from the World Health Organization as of March 15, 2021, Indonesia has 1,419,455 cumulative cases and 38,426 cumulative deaths, ranking third among countries in terms of fatalities, behind Iran and India. Because COVID-19 was disseminated through direct contact with respiratory droplets from an infected individual, it spread swiftly and widely. According to the American Centers for Disease Control and Prevention, more than 50% of transmission rates are anticipated from asymptomatic individuals. The antigen tests have an accuracy of results ranging from 80–90% and are utilized for early detection of COVID-19. The cost of the antigen test is set to increase as of September 3, 2021, with prices ranging from IDR 99.000 to IDR 109.000; however, researchers are steadfastly searching for the best alternate methods for the early diagnosis of COVID-19. According to MIT News Office, a forced cough recording can identify an asymptomatic COVID-19 infection. Through the vocal recording of a forced cough, this study uses an artificial neural network (ANN) deep learning model to identify asymptomatic COVID-19 patients. The Artificial Neural Network (ANN) can distinguish asymptomatic people from forced cough recordings with an accuracy of up to 98% and a loss value of less than 3% by employing oversampling data. This model can be applied as a free, universal method for the early identification of COVID-19 infection.

【 授权许可】

Unknown