How Indian Dermatologists are Utilizing Artificial Intelligence for Clinical Practice and Workflow Management: A Nationwide Survey with a Special Focus on atopic dermatitis

What Happened

A nationwide survey of Indian dermatologists, published on arXiv, shifts the AI-in-dermatology conversation away from the usual fixation on image-based diagnostics (e.g., melanoma detection from photos) and toward the less glamorous but arguably more impactful domain of chronic disease workflow management. The study specifically examines atopic dermatitis (AD), a chronic inflammatory skin condition that requires longitudinal tracking, treatment adjustment, and patient education—tasks that are notoriously labor-intensive and prone to inconsistency in busy clinical settings. By mapping routine clinical challenges through the lens of Indian practitioners, the researchers aim to identify where AI tools could alleviate bottlenecks, from documentation and severity scoring to follow-up scheduling and treatment adherence monitoring.

Why It Matters

This survey is significant for three reasons. First, it broadens the scope of AI application in dermatology beyond the "spot the lesion" paradigm. While convolutional neural networks for dermoscopy have dominated headlines, the real-world burden for dermatologists—especially in resource-constrained settings like India—often lies in managing chronic diseases over time. AD, with its relapsing-remitting course, requires repeated assessments, which are currently done manually using tools like the Eczema Area and Severity Index (EASI). Automating or semi-automating these workflows could free up clinician time and improve consistency.

Second, the Indian context is instructive. India has a high volume of AD patients, a diverse population with varying skin tones (which historically have been underrepresented in training datasets), and a mix of urban and rural practice settings. Any AI tool that works well in this environment would likely generalize better to other non-Western populations, addressing a known bias in dermatology AI.

Third, the survey methodology itself—asking clinicians what they actually find difficult—is a refreshing corrective to the technology-push approach that often characterizes AI development. Too many tools are built because they are technically interesting, not because they solve a pressing clinical pain point.

Implications for AI Practitioners

For AI developers and product teams, this survey offers a clear signal: stop building only diagnostic aids and start building workflow assistants. The most immediate opportunities likely lie in:

• Automated severity scoring: Using NLP to extract EASI components from clinical notes or patient-reported outcomes.
• Longitudinal tracking: AI that can summarize a patient’s disease trajectory from fragmented records.
• Personalized treatment reminders: Simple ML models that predict flare-ups based on weather, stress, or adherence data.

Practitioners should also note the importance of localization. An AI tool trained on US or European AD cohorts may fail on Indian skin types, dietary patterns, or treatment protocols. The survey underscores the need for region-specific training data and validation studies.

Key Takeaways

• The survey redirects AI focus from image-based diagnosis to chronic disease workflow management, particularly for atopic dermatitis.
• Indian dermatologists’ challenges highlight a gap in tools for longitudinal tracking, severity scoring, and adherence monitoring.
• AI practitioners should prioritize workflow automation over diagnostic novelty, and ensure models are trained on diverse, region-specific data.
• This study serves as a model for how clinician-led needs assessment can guide more practical and equitable AI development in medicine.