School Surveillance MCP Server

• 

  icon

Inspiration

I noticed that school closures happen 2-3 days before hospital data shows outbreaks. I wanted to create an early warning system using this unique signal to help healthcare teams respond faster.

What it does

I built an MCP server that exposes school-based infectious disease surveillance data as FHIR R4 resources. I implemented A2A agent coordination to analyze outbreak patterns across Japan's 47 prefectures and send automated alerts.

How we built it

I developed the MCP server with 5 FHIR-compliant tools using TypeScript and deployed it on AWS Lambda. I integrated it with Prompt Opinion's marketplace and created A2A agents for surveillance, analysis, and notification coordination.

Challenges we ran into

I struggled with FHIR R4 resource mapping for non-standard surveillance data. I also faced complexity in implementing JSON-RPC 2.0 protocol for MCP server compatibility with multiple platforms.

Accomplishments that we're proud of

I successfully deployed a production-ready system covering all 47 Japanese prefectures. I achieved full FHIR R4 compliance and demonstrated seamless A2A agent coordination with real-time outbreak detection.

What we learned

I learned how MCP and A2A protocols enable true healthcare AI interoperability. I discovered the power of FHIR standards in making surveillance data accessible to any compliant agent.

What's next for Gakkyu Alert

I plan to integrate real-time data streams from Japan's national surveillance system. I will expand A2A agent capabilities to include predictive modeling and automated public health recommendations.

Built With

• react-native