Title:SkinGPT: A Dermatology Diagnostic System with Vision Large Language Model

Abstract:Skin and subcutaneous diseases are among the major causes of the nonfatal disease burden worldwide, affecting a significant proportion of the population. However, there are three major challenges in the field of dermatology diagnosis. Firstly, there is a shortage of dermatologists available to diagnose patients. Secondly, accurately diagnosing dermatological pictures can be challenging. Lastly, providing user-friendly diagnostic reports can be difficult. Recent advancements in the field of large language models (LLMs) have shown potential for clinical applications. However, current LLMs have difficulty processing images, and there are potential privacy concerns associated with using ChatGPT's API for uploading data. In this paper, we propose SkinGPT, which is the first dermatology diagnostic system that utilizes an advanced vision-based large language model. SkinGPT is the first system of its kind, incorporating a fine-tuned version of MiniGPT-4 with a vast collection of in-house skin disease images, accompanied by doctor's notes. With SkinGPT, users can upload their own skin photos for diagnosis, and the system can autonomously determine the characteristics and categories of skin conditions, perform analysis, and provide treatment recommendations. The ability to deploy it locally and protect user privacy makes SkinGPT an attractive option for patients seeking an accurate and reliable diagnosis of their skin conditions.