MedBridge

• 

  Upper Part

• 

  Full Answer

• 

  Lower Part

Inspiration

Meet Madame Dumoulin, 81. Her health is a puzzle of doctors and pills — but the pieces don't always fit. A missed follow-up or a dangerous drug interaction can stay hidden in a fragmented system until it's too late. MedBridge was born to turn that medical silence into proactive care.

What it does

MedBridge analyzes patient FHIR records through Prompt Opinion to automatically identify care gaps across medications, appointments, and clinical flags — delivering structured, actionable reports for clinicians in seconds. No patient falls through the cracks.

How we built it

• Platform: Prompt Opinion (native A2A + FHIR)
• AI: Claude Sonnet 4 (Anthropic) for clinical reasoning + Google (AI Mode) for assistance
• Data: FHIR R4 synthetic EHR bundles
• Standards: SHARP context extension for patient data propagation

Challenges

Managing FHIR data consistency and LLM API quotas in real-time. Pivoted model architecture to Claude API for robust reasoning on complex synthetic EHR data.

What's next

Expand A2A capabilities for direct coordination with pharmacy and scheduling agents — creating a fully autonomous care continuity loop.

Built With

• a2a
• anthropic-api
• claude-sonnet-4
• fhir-r4
• google-ai-mode
• prompt-opinion