SkinMind uses graph probing for dermatology retrieval-augmented generation
Original title: SkinMind: Equipping Agentic RAG with Human-Inspired Graph Probing for Reliable Dermatology AI
Authors report average gains of 8.70 and 8.89 points with GPT-5.4-mini and Gemini-2.5-Flash on six VQA benchmarks.
SkinMind is a multimodal agentic RAG framework that lets an orchestrator probe a dermatology knowledge graph through eight hierarchical tools. SkinMind-KG was built from a 62M-token corpus of guidelines and textbooks and 423K images, yielding over 1.0M nodes and 2.0M edges. Across six dermatology VQA benchmarks and two policy-model settings, the authors report gains over direct answering and better scores than competing RAG methods.
Unconfirmed: Score gains are author-reported and not independently verified in this item.
- Tools identify entities, inspect nodes, traverse relations and paths, and retrieve text and images.
- The graph uses a dermatologist-informed schema and a conflict-resolution mechanism.
- Retrieval modality, direction, and evidence granularity are chosen during iterative probing.
- Reported gains are 8.70 points with GPT-5.4-mini and 8.89 with Gemini-2.5-Flash.
- Products
- SkinMindSkinMind-KG
- AI models
- GPT-5.4-miniGemini-2.5-Flash
From the original article
1 min read at huggingface.co
Reliable dermatology AI requires access to verifiable evidence distributed across heterogeneous textual and visual sources. However, conventional retrieval-augmented generation (RAG) typically relies on one-shot retrieval from flat corpora, limiting its ability to adapt retrieval granularity, explore multi-hop relations, and integrate multimodal clinical evidence.
By Jing Hao, Weinan Zhou, Zhihong Lu · huggingface.co
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.