Skip to content

donsirr/Wellread

Repository files navigation

WellRead: Bridging the Medical Data Silo

WellRead is a local-first health intelligence dashboard developed during Developer Camp Manila 2026. It solves the problem of medical data fragmentation by correlating unstructured personal data (Gmail, notes) with clinical metrics (PDF lab results) using the Model Context Protocol (MCP).


🚀 The Mission

The average doctor visit lasts only 15 minutes. WellRead automates the "prep work" by synthesizing a patient's symptoms and history into a clinical-grade narrative, allowing doctors to focus on treatment rather than data hunting.

🛠️ Technical Architecture

1. Model Context Protocol (MCP) Integration

WellRead functions as an MCP client that securely "bridges" local data silos.

  • Local-First Privacy: Sensitive medical context is processed directly on the user's device. Raw records are never uploaded to a centralized cloud.
  • Tooling: The system utilizes MCP tools to fetch messages from Gmail and documents from Google Drive for on-device analysis.

2. The Intelligence Engine

  • Correlation Logic: The system identifies links between unstructured symptoms (e.g., "blurred vision" in an email) and structured metrics (e.g., 8.2% HbA1c in a lab report).
  • Source Inspector: A trust layer that deep-links every AI insight to the original source text with yellow highlights to prevent hallucinations.

3. Data Schema (state.json)

The application is driven by a single Source of Truth:

  • session: Patient identity and MCP status.
  • sources: Metadata and evidence snippets for retrieved files.
  • metrics: Object containing clinical values and status labels (e.g., Blood Sugar, HbA1c).
  • activeCorrelations: Logical mapping between specific sources and metrics.
  • physicianBrief: Final S.O.A.P. note content and follow-up questions.

✨ Key Features

  • Intelligence Grid: A real-time reactive dashboard mapping life context against clinical markers.
  • Medical Time-Machine: A visualization layer that maps historical metric arrays to show health trajectories and future potential.
  • Physician Brief: A medical-grade summary designed for 30-second clinical review.
  • Demo Controller: A built-in state machine for seamless orchestration of the data lifecycle (Discovery → Analysis → Correlation → Synthesis).

👥 The Team

  • Donsir Arcilla — Lead Developer
  • Klyde Apostol — Strategy & Product
  • Atsushi Vengco — Technical Design
  • Galvin Venturina — Implementation & QA

⚖️ License & Disclosure

Built for Developer Camp Manila 2026. This is a decision-support tool, not a diagnostic one. All medical decisions should be verified by a licensed professional using the provided Source Inspector evidence.

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors