Among a host of efforts for early detection of COVID-19, a new study reveals how artificial intelligence can be useful for the same. A study carried out along with a researcher at the University of Central Florida shows that artificial intelligence can be as closely accurate to a physician in detecting COVID-19 in the lungs.
The study published in Nature Communications reveals the technique can also address some of the challenges of current testing.
According to finding of the study, AI algorithm could be trained to be able to leverage CT scans for COVID-19 pneumonia. The AI-driven CT scans detect COVID-19 pneumonia with up an accuracy of up to 90 percent, correctly identify positive cases with an accuracy of 84 percent, and negative cases with 93 accuracy.
CT Scans reliable over common tests for COVID-19 diagnosis
Meanwhile, CT scans provide detailed analysis into COVID-19 detection and advancement as compared to the commonly used reverse transcription-polymerase chain reaction tests. These tests have a high rate of false negative results, delays in processing and other challenges.
In addition, CT scans have another benefit to be able to detect COVID-19 in people without symptoms, and in those who have early symptoms, during the peak of the disease and after the symptoms are over.
Howbeit, CT scan is not always recommended for diagnosis of COVID-19. This is because the disease often mimics influenza-associated pneumonias in the scan. The new algorithm co-developed by the University of Central Florida can overcome this problem. It does so by accurate detection of COVID-19 cases, as well as differentiating them from influenza. In this way, the algorithm serves as a great potential aid for physicians.
The finding demonstrates that deep learning-built AI approach can serve as an objective and standardized instrument to assist healthcare systems and patients.