MedAssist
Emergency TriagePython / FastAPI + HTML/CSS/JS

MedAssist Review

A multi-agent emergency healthcare triage demo with FastAPI backend and static frontend.

Deployability
3/5
Value
2/5
Privacy
3/5

Each review covers deployability, value versus commercial alternatives, and privacy model. Tools that can run locally were started and exercised; mobile or backend-dependent tools were assessed from published builds, source code, and deploy guides. Ratings reflect what we were able to verify.

MedAssist — Open Source Multi-Agent Emergency Triage Demo

A multi-agent emergency healthcare triage demo with FastAPI backend and static frontend. Last updated: 2026-06-21.

One-sentence verdict: A multi-agent emergency triage demo with a clear pipeline and explainability features; illustrative only and not a validated clinical triage system.


What the System Is

MedAssist (published as MedAssist-AI) is a Python FastAPI demo for emergency triage. The deploy guide reports the backend was started on port 5020 and the static frontend on port 5021. It accepts patient information, symptoms, and vital signs, then routes the case through a multi-agent pipeline.

The documented agent stages:

  • Symptom collection
  • Risk scoring
  • Severity prediction
  • Doctor recommendation
  • Priority classification
  • Report generation
  • Community-response coordination

The backend exposes an API and serves a static HTML/CSS/JS frontend. It also includes an explainable ledger and PDF report generation.

Key data
Category Emergency Triage
Language Python / FastAPI + HTML/CSS/JS
License Unknown
Self-hosted Yes
AI provider Optional OpenAI for LLM-enhanced suggestions
Database ChromaDB
Optional NLP spaCy / scispaCy for clinical entity extraction

How to Install and Deploy

cd /data2/docker/going-global/repos/MedAssist
/root/miniconda3/envs/metagpt_env/bin/python -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt

# Optional: enable clinical NER
python -m spacy download en_core_web_sm

# Terminal 1 — backend
uvicorn main:app --host 0.0.0.0 --port 5020

# Terminal 2 — frontend
python -m http.server 5021

Then open http://localhost:5021/index.html. The frontend JavaScript expects the API at http://localhost:5020; the deploy guide notes this was patched from the original http://localhost:8000.


How to Test

The documented test flow is:

  1. Start the backend on port 5020 and the frontend on port 5021.
  2. Open http://localhost:5021/index.html.
  3. Enter patient information, symptoms, and vital signs.
  4. Submit the case and observe the agent execution ledger.
  5. Verify the generated triage report and priority classification.
  6. Optional: run python scripts/test_triage.py to validate API behavior.

The README states the triage outputs are illustrative only.


Privacy & Compliance

MedAssist is not HIPAA compliant. It stores data in ChromaDB and optionally calls OpenAI. There is no authentication, encryption, or access control described in the deploy guide. Do not use it for real patient data.


MedAssist vs Commercial Triage Systems

Dimension MedAssist Commercial Systems (e.g., Epic, eVisit, Babylon Health)
Cost Free / self-hosted Enterprise licensing or per-use
Data model Local backend + optional OpenAI Cloud-hosted, EHR-integrated
Clinical validation None; demo outputs Medically reviewed, often regulated
Setup effort Two-terminal Python setup Vendor-managed
Explainability Agent ledger visible in UI Varies by product
Customizability Full source code Limited configuration

Who Should Use It

  • Developers building multi-agent triage or emergency-response demos.
  • Teams evaluating FastAPI + ChromaDB + static-frontend architectures.
  • Hackathon projects that need a visual agent pipeline.

Who Shouldn't Use It

  • Any real emergency or clinical triage workflow.
  • Organizations needing HIPAA-compliant patient handling.
  • Users expecting a validated risk-score or medical advice.

FAQ

Is MedAssist a real emergency triage system?

No. The README and deploy guide describe it as a demo with illustrative outputs that are not medical advice.

Does it require OpenAI?

OpenAI is optional for LLM-enhanced suggestions. The demo can run without it, using rule-based or local logic.

How is the frontend served?

The frontend is a set of static HTML/CSS/JS files served by Python's http.server. It communicates with the FastAPI backend on a separate port.


Verdict

MedAssist is a readable multi-agent triage demo with a clear frontend/backend split and an explainable ledger. It is useful for teaching and prototyping, but it is not a substitute for validated emergency triage or clinical decision support.

Ratings: Deployability 3/5 · Value vs Commercial 2/5 · Privacy Compliance 3/5