Google has recently announced that its Lens image search feature can now help identify skin conditions. The technology, integrated into Google apps on both iOS and Android, allows users to “search for skin conditions” such as an unusual mole or rash by simply taking a picture or uploading a photo through Lens. The tool provides visual matches to inform users about potential skin conditions. This could be useful if you want more information about a bump on your lip, a line on a nail, or hair loss from your scalp. Google emphasizes that the results are meant to be informational only and not a diagnosis, advising users to consult their medical authority for advice.
Google has been leveraging AI image recognition to identify skin conditions for some years now. In 2021, the company unveiled a tool at its developer conference designed to identify skin, hair, and nail conditions using a combination of photos and survey responses. At the time, Google claimed the tool could recognize 288 different conditions and would present the correct condition in the top three suggestions 84 percent of the time. Google’s DermAssist tool, which is currently being market-tested through a limited release, is CE-marked as a Class 1 Medical Device in the European Economic Area. However, it has not yet been evaluated by the United States FDA and is intended only for informational purposes, not as a medical diagnosis.
Despite these advancements, caution is advised with AI diagnostic tools, as these systems have shown less accuracy for users with darker skin tones. The research highlighted a lack of skin type category data across many freely available image databases used to train AI systems and a lack of images of dark-skinned individuals in databases that did include this information.
Google has actively been addressing these issues. The tech giant partnered with Harvard professor Ellis Monk last year to promote his Monk Skin Tone Scale (MST) and outline best practices for its application in AI development. Google’s teams have been using the MST Scale for machine learning labeling tasks, ethnographic research, and fairness testing1. Google’s deep learning system was also reported to have an accuracy rate of 87.9 percent for Black patients, which was higher than for other ethnicities in 2021.