AI skin doctor: The All-India Institute of Medical Sciences is working with Google to create an artificial intelligence application that could help diagnose skin conditions, particularly in areas where dermatologists are scarce, reported The Indian Express.
The app, called Indus Derma, is scheduled to launch next year. Users will be able to photograph a skin rash or other concern, upload the image along with basic information like age and symptoms, and receive a potential diagnosis generated by AI.
The technology uses computer vision and large language models to analyse photos, compare them against databases of skin conditions, and suggest what the problem might be. The app will also indicate whether someone should see a dermatologist urgently, can manage the condition locally, or needs additional testing.
Also Read | Vitamin B3 supplement tied to lower skin cancer risk: Study
Dr Somesh Gupta, a dermatology professor at AIIMS who is involved with the project, said access to trained dermatologists remains limited outside cities. Most dermatologists in India practice in urban areas, and many focus on cosmetic work rather than general dermatology. This leaves people in smaller towns and rural areas relying on general physicians or other providers with less specialised training.
Dermatology problems rank among the most searched health topics online globally. But the gap between demand and access to specialists prompted the collaboration between AIIMS and Google. The app will not generate legally valid prescriptions. Any medication recommendations from the AI must be reviewed and approved by a qualified medical professional. “This is a decision support tool, not a replacement for doctors,” Gupta said.
One major challenge with existing global AI dermatology tools is that they were trained mostly on images of white skin. This creates accuracy problems when diagnosing patients with different skin tones. AIIMS and Google are developing foundation models specifically for India that will retrain global AI systems using images of Indian skin tones and conditions common in the region. “Skin colour significantly influences visual diagnosis. That’s why local data is essential,” Gupta told The Indian Express.
The project is still in development and validation. It needs regulatory approval, accuracy testing and extensive trials before public release. The team aims for accuracy between 80 and 90 percent, well above the roughly 40 percent accuracy rate seen when non-specialists manage skin conditions.
The app targets general physicians, healthcare workers and patients in areas with limited specialist access rather than experienced dermatologists. It will help users gauge the seriousness of their condition and learn basic care measures. For patients, the app will provide explanations about what a condition is, why it occurs and general management steps. Doctors can review the AI suggestions and incorporate them into clinical decisions where appropriate.