Japanese Researchers Develop AI Model to Assess Eczema Severity
A team of researchers from Keio University School of Medicine, Kyoto Prefectural University of Medicine, and Teikyo University, in partnership with Atopiyo LLC, has introduced a new AI model that can objectively assess eczema severity using patient-uploaded smartphone images. The study, recently published in Allergy, highlights how artificial intelligence can transform dermatological care.
Atopic dermatitis (AD), a chronic skin condition, often requires long-term monitoring. Traditional evaluations rely heavily on patient-reported symptoms like itch and sleep loss, which don’t always match visible inflammation. This mismatch creates a need for standardized, objective tools—precisely where AI offers value.
Leveraging Data from Japan’s Largest Eczema Platform
To develop the model, researchers utilized data from Atopiyo, Japan’s largest AD support platform. Since 2018, more than 28,000 users have uploaded over 57,000 images and symptom logs. The AI system combines three core functions: body part detection, eczema lesion identification, and a severity scoring algorithm based on the Three Item Severity (TIS) scale. This scale assesses redness, swelling, and excoriation.
Also read: OpenAI Launches Codex, an AI Coding Agent
Trained on 880 images with self-reported itch scores, the model achieved high diagnostic performance. In validation testing with 220 images, AI-generated scores closely matched dermatologist ratings, demonstrating strong accuracy. The AI-TIS score also aligned with the objective SCORAD index, further validating the tool’s potential clinical utility.
Bridging the Gap Between Symptoms and Skin Appearance
Interestingly, the study revealed only a weak correlation between self-reported itch and AI-assessed severity. This finding emphasizes the need for digital biomarkers to help accurately measure disease activity and guide treatment decisions.
Looking ahead, the researchers plan to refine the model further by including wider age ranges, diverse skin tones, and additional clinical scoring systems like EASI and SCORAD. Their long-term goal is to support real-world teledermatology applications, helping patients and clinicians monitor eczema remotely and accurately.
This development marks a step forward in AI-powered healthcare, offering a scalable and cost-effective solution for chronic skin disease management.
