Mole or cancer? The algorithm that gets one in three melanomas wrong and erases patients with dark skin
Mole or cancer? The algorithm that gets one in three melanomas wrong and erases patients with dark skin

civio.es
Mole or cancer? The algorithm that gets one in three melanomas wrong and erases patients with dark skin

The Basque Country is implementing Quantus Skin in its health clinics after an investment of 1.6 million euros. Specialists criticise the artificial intelligence developed by the Asisa subsidiary due to its "poor” and “dangerous" results. The algorithm has been trained only with data from white patients.
Yeah, a 31% false negative rate is completely unacceptable. If it was a very low false negative rate and a 31% false positive rate I could see some value there, but this is garbage.
Which is sad because AI does have real diagnostic potential, but obviously diagnosing from a picture isn't included in that potential.
But honestly, that's to be expected. There are so many variables that go into taking a picture of a small spot n a 3D surface that it's stupid to use that as an input. From the article:
If we were talking about AI reading biopsy slides, I'd bet that the results would be much different.