AI will provide more accurate diagnoses of dermatitis

NEW YORK (HealthDay News)—Eczema sufferers may soon find it easier to keep track of their skin condition, thanks to a newly developed AI that can assess its severity using images uploaded from cellphones.

AI demonstrated high diagnostic accuracy when evaluating eczema displayed in photos of symptoms uploaded by patients using their cellphone cameras, researchers reported in the journal Allergy.

“Many patients with eczema have difficulty assessing the severity of their disease on their own,” said senior researcher Dr. Takeya Adachi, assistant professor of dermatology at Keio University School of Medicine in Tokyo.

“Our AI model enables objective, real-time monitoring using only a cell phone, empowering patients and potentially improving disease management.”

Eczema, also known as atopic dermatitis, tends to flare repeatedly, requiring long-term monitoring and treatment adjustments, the researchers noted in background notes.

However, symptoms reported by patients, such as itching or lack of sleep, do not always align with how eczema rashes appear to the naked eye.

Popular application

For the new AI, the researchers collected data from a popular eczema-tracking app called Atopiyo in Japan. More than 28,000 users have shared nearly 57,000 photos of symptoms and personal comments since 2018.

The team developed AI to assess the severity of eczema based on its location, the size of the rash, and the resulting redness, swelling, or irritation caused by scratching the lesion.

The artificial intelligence was trained on a set of 880 images with self-reported itch scores, before researchers tested its effectiveness on another set of 220 test images.

In the test, the AI ​​correctly detected 98 percent of body parts and 100 percent of eczema areas, and its analysis correlated well with severity scores from board-certified dermatologists and allergists, the researchers said.

More training

Next, the scientists plan to further train the AI ​​by incorporating more skin types and age ranges, and including additional features from other clinical scoring systems for eczema.

"The AI ​​model developed in this study has the potential to help eczema patients objectively assess their skin condition, facilitating timely and appropriate treatment," the experts noted in their article.

“This study lays the groundwork for future advancements in AI-powered dermatological evaluations, improving both patient care and clinical research.”