Ian Patrick, FISM News
A new study published in Nature Communications reveals that AI can be used in diagnosing COVID-19 in a person’s lungs through CT (computed tomography) scans. The study says that using AI in chest CT scans “has the potential to aid in rapid evaluation…for differentiation of COVID-19 findings from other clinical entities.”
The study used the scans of 1,280 patients to train the algorithms in locating and identifying the COVID pneumonia. The CT scans used were from a diverse set of individuals in order to improve accuracy in the AI. Once the trained algorithm was applied to independent scans of 1,337 patients, the AI was able to detect the COVID-19 pneumonia 90.8% of the time. The AI achieved “84% sensitivity and 93% specificity” from within the grander accuracy percentage.
The AI performed well but still made mistakes, as the researchers tested. Part of the experiment used scans from 140 patients with laboratory-confirmed non-COVID pneumonia. From this set, the AI incorrectly assessed that 10% of this population had COVID.
Despite this last number, the researchers report that the AI had “acceptable performance metrics.” Overall, the researchers think that the AI can “readily identify CT scans with COVID-19 associated pneumonia, as well as distinguish non-COVID related pneumonias with high specificity in diverse patient populations.”
Sourced from Nature Communications