Inspiration
Managing medical inventories in hospitals is often slow, error-prone, and prone to shortages or overstocking. We wanted to create a system that not only tracks and secures medical supplies but also communicates naturally with healthcare staff through AI chatbots and voice assistants—making the process faster, smarter, and more accessible.
What it does
Spider (MedChain) is a blockchain-powered medical inventory management system enhanced with AI-driven chatbots and voice models. It:
- Tracks medicines and supplies in real-time using RFID tags
- Uses ML models to forecast demand and prevent shortages
- Offers natural conversation interfaces (chat & voice) for stock queries, ordering, and alerts
- Provides secure, immutable records on blockchain for trust and compliance
- Integrates AI diagnostics like skin cancer detection for proactive patient care
How we built it
- Frontend: React + TypeScript + TailwindCSS + shadcn/ui for a fast, responsive UI
- Backend: FastAPI with PostgreSQL & Redis for high-speed API responses
- Blockchain: Solidity smart contracts deployed on Ethereum testnet with Ethers.js integration
- AI/ML Models:
- Demand forecasting (Random Forest, Gradient Boosting, LSTM)
- NLP-powered chatbot with context retention
- Speech-to-text and text-to-speech voice assistant
- CNN for skin cancer detection
- Demand forecasting (Random Forest, Gradient Boosting, LSTM)
- RFID: Live inventory tracking integrated with backend APIs
Challenges we ran into
- Integrating blockchain smart contracts seamlessly with the backend
- Achieving real-time voice processing without latency
- Training ML models with limited medical datasets
- Ensuring HIPAA-compliant security while maintaining performance
- Making the chatbot context-aware across multiple conversation sessions
Accomplishments that we're proud of
- Built a fully functional blockchain + AI + voice-enabled medical inventory system within hackathon time
- Deployed smart contracts with secure supplier verification
- Created a chatbot that understands medical inventory queries in plain English
- Integrated RFID for real-time stock visibility
- Developed a CNN model for skin cancer detection with over 89% accuracy
What we learned
- How to combine blockchain transparency with AI usability in healthcare
- Advanced RFID integration for live asset tracking
- Training and deploying ML models for both prediction and diagnostics
- Optimizing voice AI for real-time response in a high-stakes domain
What's next for Spider
- Add multi-language support for chatbot & voice assistant
- Integrate with national drug registries for authenticity verification
- Deploy on a private blockchain for hospitals and clinics
- Expand AI diagnostics to other medical conditions
- Partner with healthcare providers for real-world pilot testing
Built With
- ethers.js
- fastapi
- keras
- opencv
- postgresql
- python
- react
- redis
- rfid
- scikit-learn
- shadcn/ui
- solidity
- tailwindcss
- technology
- tensorflow
- typescript
Log in or sign up for Devpost to join the conversation.