AI’s results in its real-world application in radiology and other medical fields have often fallen far short of expectations, but delving deeper into the reasons behind such letdowns could hold the key to future breakthroughs, even if AI ends up playing a more supportive than dominant role, argues data scientist Dan Elton.

As the novel coronavirus evolved from epidemic to pandemic, early research suggested CT scans provided the best diagnosis for COVID-19. It turns out that the information was flawed. Data scientist and Google Scholar Dan Elton takes a closer look at the studies published.