Back to index
//PROJECT DOSSIER

Context-aware Data-Driven Medical Assistant

LangChain
Neo4j
MCP
Django
Next.js
TypeScript
Detailed screenshot of Context-aware Data-Driven Medical Assistant project showing the interface and key features
OVERVIEW

Project Overview

A clinician-facing assistant that answers natural-language questions over an Electronic Health Record knowledge graph, so staff find records without searching by hand. A LangChain agent reaches Neo4j through a Model Context Protocol server, querying data via one audited Cypher tool rather than holding the schema or credentials. Patients, appointments, diagnoses, prescriptions and vitals are modelled as a graph behind a Django Ninja API and a Next.js chat interface.

ARCHITECTURE

System Architecture

Architecture diagram for Context-aware Data-Driven Medical Assistant showing system components and their interactions
KEY FEATURES

Key Features

  • 01Multi-chat UI with speech-to-text capabilities and markdown-formatted replies
  • 02Seven domain modes (vitals, labs, etc.) with automatic prompt slice loading
  • 03Structured-chat agent guaranteeing JSON tool calls with retry mechanisms for parse errors
  • 04Live Neo4j query execution with safety rules and lower-case literal enforcement
  • 05FastMCP transport layer for efficient communication between components

Command Palette

Navigate the site, open links, or toggle the theme. Use arrow keys to move, Enter to select, Escape to close.