"I created PatientSync AI because I wanted to rise to the challenge of building a secure, interoperable 'Superhero' agent for the Agents Assemble: The Healthcare AI Endgame Hackathon!" 🚀
💡 Inspiration
I wanted to solve "The Last Mile" problem in healthcare. 🏥 Doctors often have to spend hours digging through messy, technical medical records. I was inspired to build a tool that bridges the gap between complex data and the humans who need it most, giving clinicians more time to focus on their patients. 🤝
🤖 What it does
PatientSync AI is an intelligent assistant that acts as a secure bridge for healthcare data:
📋 Smart Fetch: It securely accesses patient records using FHIR standards.
✍️ Plain English: It translates difficult medical jargon into simple summaries.
💊 Data Extraction: It instantly pulls out medications, allergies, and diagnoses.
🛡️ Safety First: It has built-in triage logic to advise users to see a doctor for high-risk symptoms.
🏗️ How I built it
I used n8n as the "Brain" to orchestrate my AI Superpowers:
🧠 Groq & Gemini: For lightning-fast reasoning and natural conversation.
🔗 MCP (Model Context Protocol): To ensure the agent can talk to any healthcare system.
🔍 Cohere & Qdrant: For high-accuracy medical search and context.
💻 JavaScript: For custom data cleaning to ensure the AI always stays accurate.
🚧 Challenges I ran into
One of the biggest hurdles was handling Data Consistency. 🛠️ I faced a technical error where the AI expected a "String" but received an "Object" from the FHIR bundle. I solved this by building a custom data-cleaning step in n8n to ensure the AI only receives the clean text it needs to process safely.
Built With
- cohere
- fhir-standards
- gemini-api
- groq
- javascript
- n8n
- qdrant
Log in or sign up for Devpost to join the conversation.