DermaScribe AI: AI-Powered Backend for Dermatology Workflows

*Using Spring Boot, JWT, and OpenAI to enhance medical transcription*

🩺 Introduction

In clinical settings, dermatologists often spend a significant amount of time transcribing session notes and summarizing case details. What if AI could assist with this routine work—helping doctors focus more on diagnosis and treatment?

DermaScribe AI is a backend service built with Spring Boot that integrates AI models (like OpenAI) to support dermatologists through automated transcription and analysis. This system securely handles session data, processes it intelligently, and exposes clean REST APIs for seamless integration with frontend or mobile apps.

🔍 Why DermaScribe AI?

Medical transcription is time-consuming, repetitive, and prone to inconsistency. By integrating AI into clinical backend systems, we can:

Automate session summarization
Provide intelligent suggestions for diagnosis
Offer structured session storage and easy retrieval

🧱 System Architecture

The project is built using a layered architecture, making it modular and extendable:

✅ Core Features:

JWT Authentication & Authorization using Spring Security
AI-Powered Analysis via OpenAI's API
RESTful APIs for session management
Modular Design for maintainability and scalability

🗂 Layered Breakdown:

Security Layer
Handles authentication/authorization with SecurityConfig, JwtTokenProvider, and a customizable CustomUserDetailsService.
Controller Layer

AuthController: Manages user login and token generation
SessionController: Exposes endpoints to create and retrieve dermatology sessions

Service Layer

AIProcessorService: Central logic to call and process AI output
OpenAIService: Makes requests to OpenAI and parses responses

Data Layer

JPA repositories and domain models to persist dermatology sessions

🧪 Running the Project

🔧 Prerequisites:

Java 17+
Maven 3.6+
OpenAI API Key
Git

📦 Clone the Repository:

git clone https://github.com/amkumar072/DermaScribe-AI
cd derma_scribe_ai_springboot

⚙️ Setup Configuration:

Update application.properties with:

OpenAI API key
JWT secrets
Database settings (or use in-memory H2 for dev)

🔨 Build and Run:

mvn clean install
mvn spring-boot:run

By default, the server will run on http://localhost:8080.

🧪 API Testing:

Use Postman or Swagger UI to test:

/auth/login
/api/sessions/create
/api/sessions/{id}
/api/{id}/upload

Sample credentials (dev mode):

Username: dermatologist
Password: password

📌 What’s Next?

Some ideas for future enhancements:

NLP fine-tuning for more specialized medical suggestions
Audit logging and analytics dashboard
Frontend (React or Android) to consume APIs

🚀 Conclusion

DermaScribe AI is a step toward smarter, AI-assisted healthcare. It’s modular, secure, and designed for real-world extensibility. If you're a developer, clinician, or AI enthusiast, I’d love your feedback or collaboration!

🔗 GitHub: https://github.com/amkumar072/DermaScribe-AI