Project: Creative Apps - MedViva AI: The Clinical Reasoning Training System

[MiniTheCoder]

Track

Creative Apps (GitHub Copilot)

Project Name

MedViva AI: The Clinical Reasoning Training System

GitHub Username

MiniTheCoder

Repository URL

https://github.com/MiniTheCoder/medviva---ai

Project Description

MedViva AI turns any medical textbook into a personal Socratic AI examiner for NEET-PG students and MBBS viva preparation.

Students upload their PDF textbook into one of 19 subject tabs. Azure AI Search indexes it into semantic vectors. The AI then conducts a rigorous oral viva — asking clinical scenarios, evaluating answers, grading them (Correct/Partial/Incorrect), and always demanding mechanism-level reasoning before moving forward.

Key Features:

• Dual-Mode: Viva Mode (Socratic dialogue) and MCQ Mode (board-style clinical vignettes with realistic distractors)
• Two-Tier Clinical Safety Guardrail: Cross-checks AI responses against global medical consensus. Life-threatening errors trigger a persistent ⚠️ CLINICAL SAFETY NOTICE with mandatory "Mark as Reviewed with Faculty" acknowledgement — creating a medico-legal audit trail
• Board Readiness Index: Tracks accuracy with formula: True Readiness = Student Accuracy × Document Reliability
• Session Isolation: Each subject tab maintains independent knowledge context
• Saved Questions Bank: Bookmark high-yield questions for revision

Built with Azure AI Foundry (gpt-5.1 + text-embedding-3-small), Azure AI Search (HNSW Hybrid index), Next.js 16, and GitHub Copilot for development assistance.

Demo Video or Screenshots

Demo Video: https://www.youtube.com/watch?v=Nu2vAxWR8dU

Primary Programming Language

TypeScript/JavaScript

Key Technologies Used

• Azure AI Foundry (Azure OpenAI gpt-5.1)
• Azure AI Search (HNSW + Hybrid Semantic Index)
• text-embedding-3-small for vectorization
• Next.js 16 (App Router) + React
• GitHub Copilot (Inline + Chat)
• unpdf (serverless PDF parsing)

Submission Type

Individual

Team Members

No response

Submission Requirements

• My project meets the track-specific challenge requirements
• My repository includes a comprehensive README.md with setup instructions
• My code does not contain hardcoded API keys or secrets
• I have included demo materials (video or screenshots)
• My project is my own work with proper attribution for any third-party code
• I agree to the Code of Conduct
• I have read and agree to the Disclaimer
• My submission does NOT contain any confidential, proprietary, or sensitive information
• I confirm I have the rights to submit this content and grant the necessary licenses

Quick Setup Summary

1. Clone the repo: git clone https://github.com/MiniTheCoder/medviva---ai
2. Install dependencies: npm install
3. Copy .env.example to .env.local and fill in your Azure credentials
4. Run: npm run dev
5. Open http://localhost:3000/viva

Technical Highlights

• Two-Tier Clinical Safety Guardrail: Detects dangerous errors IN the student's own textbook — a problem no other RAG study tool addresses
• Medico-legal audit trail with un-dismissable "Mark as Reviewed with Faculty" banner
• True Readiness Formula: Student Accuracy × Document Reliability
• Strict RAG grounding with per-session document isolation across 19 NEET-PG subject tabs
• Socratic State Machine: Forces mechanism-level reasoning, not just answer recall

Challenges & Learnings

The hardest challenge was building the Two-Tier Safety system — detecting when a student's own textbook contains dangerous medical errors, without breaking exam flow. The solution was a Clinical Consensus Override layer in the system prompt that cross-checks retrieved context against foundational medical knowledge before responding, appending the appropriate warning tier based on severity.

Contact Information

miniagg06@gmail.com

Country/Region

India