MedFlowAssist

Overview

MedFlowAssist is a multi-agent medical assistant that combines the power of MedGemma, medical speech recognition (MedASR), and LiteLLM to deliver clinical-grade AI assistance across a wide range of healthcare workflows.

Key Features

• 13 clinical workflows covering documentation, triage, diagnosis support, and more
• 12 pre-built scenarios ready to deploy out of the box
• Voice dictation via MedASR for hands-free clinical note-taking
• Multimodal input — image, PDF, and CSV upload supported
• Multi-agent orchestration with LangGraph-style task delegation

Clinical Workflows

The assistant covers end-to-end workflows including:

• Patient history summarisation
• Differential diagnosis generation
• Radiology report analysis
• Lab result interpretation
• Clinical note drafting (SOAP format)
• Drug interaction checking
• ICD-10 coding assistance

Architecture

User Input (text / voice / image / PDF)
         │
         ▼
   MedFlowAssist Router
    ├── MedGemma (vision + language)
    ├── MedASR (speech → text)
    └── LiteLLM Proxy (model routing)
         │
         ▼
   Clinical Workflow Agents
    ├── Summariser
    ├── Diagnostics
    ├── Documentation
    └── ...

Quick Start

git clone https://github.com/YouvenZ/MedFlowAssit
cd MedFlowAssit
pip install -r requirements.txt
python app.py

Visit http://localhost:5000 to access the interface.

Prerequisites

• Python 3.11+
• API key for MedGemma or compatible model endpoint
• (Optional) MedASR setup for voice features