DiagnoBot | Devpost

🧠 Inspiration

The global healthcare system is under immense pressure—patients often experience long wait times, overwhelmed emergency rooms, and misrouted care. Our team was inspired by the idea that AI agents can help bridge the efficiency gap in diagnostics and triage. We envisioned a platform where non-critical cases are handled instantly by AI, while critical ones are routed to the right medical professionals faster, ensuring smarter, safer, and more accessible healthcare.

🤖 What it does

DiagnoBot is an AI-powered healthcare portal that automates triage, symptom analysis, and patient routing based on urgency.

Key Features:

Symptom Checker: Voice or text input with AI urgency detection
Smart Diagnosis (for non-urgent cases): Instant LLM-generated diagnosis and care tips
Live Doctor Consultation: For semi-urgent cases, video/chat room connects patients with licensed doctors
Doctor and Medical User Portal: Manage schedules, verify credentials, consult patients, and generate reports
Patient Dashboard: View medical history, reports, and appointments
AI Report Generation: Each consultation generates a shareable report
Payment System: Secure appointment payments and history tracking

🛠️ How we built it

DiagnoBot was engineered as a full-stack AI-driven health platform using modern web technologies, LLM tooling, and cloud infrastructure. Here's a breakdown:

💻 Frontend

React.js + TailwindCSS: Built dynamic, responsive UI with clean, accessible design and styling.

🔧 Backend

Python + FastAPI: High-performance backend APIs and endpoints
LangChain + RAG (Retrieval-Augmented Generation): Context-aware response generation from LLMs
SQLAlchemy: Database ORM for efficient and scalable data models
OAuth: Role-based authentication and secure login flows

🧠 AI & NLP

OpenAI + Gemini API: Smart diagnosis, care tips, and report generation
Whisper API: Voice-to-text transcription for symptom input
Pandas: Data analyzation, structuring and medical history management

🗃️ Database & Storage

MongoDB Atlas: Stores user records, diagnosis results, and consultation data
Vector Store: Enables RAG to retrieve relevant context for LLMs
SQLAlchemy Models: Additional structured storage logic for report handling

☁️ Deployment & DevOps

Google Cloud Platform (GCP): Hosting, environment management, and scaling
Docker: Containerized environment for local development and cloud deployment

We divided tasks based on skillsets:

Joanna led machine learning and RAG integration
Ewa built the responsive frontend and component design
Vrushanki & Akanksh handled backend logic, endpoints, and database models

🧗‍♀️ Challenges we ran into

Data Alignment: Mapping symptom inputs to LLM-friendly formats required building a reliable preprocessing flow.
Urgency Detection: Creating a logic that separates Level 2 vs Level 3 triage cases without human bias took iteration.
Real-Time Communication: Setting up secure and seamless chat/video sessions with proper data flow between user and doctor dashboards.
Component Overload: Designing the UI to feel intuitive and not overwhelming with so many moving parts.
Time Constraints: Juggling multiple integrations (LLM, Whisper, Payments) in a short hackathon window.

🏆 Accomplishments that we're proud of

Created a working triage flow that simulates real-world healthcare decision-making
Implemented LLM-backed diagnosis and generated structured medical reports
Integrated voice-to-text input, video consultation, and a multi-role dashboard system
Designed a scalable architecture that could realistically support healthcare orgs

📚 What we learned

How to build multi-agent systems with real-world applications
Practical Langchain + RAG workflows for retrieving medical answers responsibly
The importance of UI/UX in healthcare tech, especially for users under stress
Building role-based dashboards and authentication logic in real-world apps
How to plan and execute an AI-driven product collaboratively under pressure

🚀 What's next for DiagnoBot

Level 1 Emergency Routing: Future plans to escalate emergencies to EMS services
Phone Bot Integration: Add AI voice assistant to handle inbound calls
Doctor Recommendation Engine: Match patients to doctors based on symptoms, specialty, and availability
Health Insights: Track user trends and suggest preventive care
HIPAA Compliance & Security: Strengthen data protection for real deployments
Clinical Trials / NGO Partnerships: Validate and pilot in rural or under-resourced areas
Payment System Integeration: Integerate secure appointment payments and history tracking (Stripe for payments)