How should artificial intelligence screen for skin cancer?
In dermatology, the past year and so have been marked by breakthroughs in machine learning. This has made automated diagnosis of some cutaneous lesions achievable, comments an article in JAMA Dermatology by authors George A. Zakhem, BS; Catherine C. Motosko, BS; Roger S. Ho, MD, MS, MPH.
Although machine learning systems have been around since the 1950s, the past decade has seen an unprecedented surge in their adoption. This rise in artificial intelligence can be attributed to an explosion in the generation of digital data, improvements in algorithms, and substantial advancements in computing hardware. As building these systems becomes increasingly routine, they are now used on nearly every technological platform we use daily and underlie a range of functions, including speech transcription, recommender systems, and image recognition.
Algorithms for classification
Algorithms have been developed that approach dermatologist-level classification of cutaneous tumours. These technologies show tremendous promise for the improvement of skin cancer screening. Still, many aspects of their use have yet to be elucidated.
One of the models developed by Han et al has been optimized for mobile use and can be accessed without a subscription or login (http://modelderm.com/) - the photo above. But, it is still unclear whether the site is designed for patients, nonspecialist clinicians or dermatologists.