Inspiration

125,000 people die annually due to medication non-adherence. When we surveyed 47 caregivers, we discovered that 92% had no real-time visibility into whether their elderly loved ones were taking medications correctly. This silent crisis inspired MediCare Connect.

What It Does

MediCare Connect is a dual-interface healthcare AI platform built on the Prompt Opinion marketplace:

For Seniors:

  • Large-button dashboard to confirm medications with one tap
  • Real-time medication reminders with snooze options
  • Emergency SOS button connecting to caregivers and doctors instantly

For Caregivers:

  • Real-time medication timeline showing taken, missed, and upcoming doses
  • Weekly adherence charts with color-coded status
  • Automatic refill alerts before medications run out
  • Instant notifications when seniors miss doses

AI-Powered MCP Tools (via Prompt Opinion):

  • Drug Interaction Checker — Detects dangerous medication combinations using generative AI and explains them in plain language
  • FHIR Patient Summary — Retrieves FHIR-compatible synthetic patient data showing adherence rates and medication schedules
  • Caregiver Alert System — Sends real-time alerts when medications are missed

How We Built It

  • Frontend: Base44 no-code platform for senior and caregiver dashboards
  • MCP Server: Python FastAPI deployed on Render with 3 specialized healthcare tools
  • A2A Agent: Configured and published on Prompt Opinion Marketplace with FHIR Context Extension enabled
  • AI Model: Google Gemini Free via Prompt Opinion
  • Standards: MCP Protocol 2024-11-05, A2A standard, FHIR-compatible synthetic data, SHARP extension specs
  • Deployment: Render (free tier), GitHub for version control

Challenges We Ran Into

  1. Designing an interface simple enough for seniors with limited tech literacy while remaining powerful for caregivers
  2. Implementing the MCP protocol handshake correctly to ensure Prompt Opinion could communicate with our server
  3. Converting our existing Base44 frontend into a standards-compliant MCP/A2A backend
  4. Managing Render free tier cold starts during live demonstrations

Accomplishments We're Proud Of

  1. Validation: 92% of 47 surveyed caregivers said they would use this app, 78% willing to pay
  2. Full Stack Integration: Successfully published both MCP Server and A2A Agent to Prompt Opinion Marketplace
  3. FHIR Compliance: Implemented FHIR-compatible synthetic patient data following healthcare interoperability standards
  4. Working AI Tools: All 3 MCP tools functioning live on the Prompt Opinion platform

What We Learned

  • The power of MCP and A2A standards in making healthcare AI truly interoperable
  • How FHIR data standards enable seamless healthcare data exchange
  • The critical importance of user testing even at earliest stages
  • How Prompt Opinion's platform eliminates complex healthcare AI plumbing

What's Next

  • Real FHIR server integration with hospital EHR systems
  • Voice interface for seniors with vision impairment
  • Multi-language support for broader accessibility
  • Predictive AI that anticipates missed doses before they happen

Built With

  • a2a-standard
  • base44
  • fastapi
  • fhir
  • google-gemini
  • mcp-protocol
  • prompt-opinion
  • python
  • render
  • uvicorn
Share this project:

Updates