In a press release, the educated community views the provision of such tools without being transparent about their performance for rare but deadly cancers as “ethically questionable.”
These findings come from a study led by Lloyd Still of the Royal London Hospital (Barts Health NHS Trust), Queen Mary University of London, and colleagues. In their summary, they mention that machine learning models have been developed to diagnose skin cancers in studies that report performance similar to that of dermatologists.
Skin lesions are very common and anyone who discovers them wonders if they could be cancerous. For the general public, these algorithms available through mobile applications can be useful. The researcher explained during a press conference that the detection of rare but aggressive cancers is an important interest because they are not sufficiently present in the training databases used to develop these algorithms.
Among these rare types of cancer,
To access the rest of the article, you must accept performance and tracking cookies. It allows us to create statistics that we share with our advertisers without revealing our identity. Thanks to advertising, this site is funded and the articles written by our journalists are available to you for free.
To modify your cookie settings, Press here.
