https://github.com/geobatpo07/sekou-ai

https://github.com/geobatpo07/sekou-ai

Last synced: 8 days ago
JSON representation

• Host: GitHub
• URL: https://github.com/geobatpo07/sekou-ai
• Owner: Geobatpo07
• License: mit
• Created: 2025-07-25T19:21:47.000Z (3 months ago)
• Default Branch: main
• Last Pushed: 2025-08-14T21:38:00.000Z (2 months ago)
• Last Synced: 2025-08-29T03:55:50.698Z (about 2 months ago)
• Language: HTML
• Size: 172 KB
• Stars: 0
• Watchers: 0
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

          
# sekou-ai

Production-ready MVP of SekouAI — a healthcare AI for intelligent medical triage.

- Backend: FastAPI (Python) with SQLite (app.db) via SQLAlchemy

- Frontend: Responsive HTML/CSS/JS designed for clinical environments

- Risk levels: "low", "medium", "high"

- All predictions are persisted with inputs and timestamp for traceability

## Run locally

1) Create/activate a Python 3.12+ environment

2) Install deps:

   - pip install -r requirements.txt

3) Start API:

   - uvicorn backend.main:app --host 0.0.0.0 --port 8000

4) Open the app:

   - Visit http://localhost:8000/ (served by FastAPI templates)

   - Optional routes: http://localhost:8000/add, http://localhost:8000/history, http://localhost:8000/patients

### Frontend features

- Dashboard with lateral sidebar navigation (Dashboard, Add Patient, History).

- Risk analysis dashboard: total counts and low/medium/high distribution with a simple bar chart.

- Multiple patient forms: Adult triage, Pediatric template quick actions, and optional custom vitals.

- Predict buttons open a dedicated risk overlay with status and metadata.

- Medical templates to pre-fill triage form: Adult Respiratory Distress, Adult Mild Symptoms, Pediatric Fever & Cough.

- Accessible form with labels, fieldset/legend, aria-live results.

- Recent predictions table with risk badges and timestamps.

- Separated JavaScript modules: static/js/{api,utils,dashboard,forms,app}.js.

- Patients list and patient detail/edit/delete views (backed by persisted prediction records).

- Legacy predict demo kept for backward compatibility/testing.

## API endpoints

- GET http://localhost:8000/health

- POST http://localhost:8000/triage

  Body example:

  {

    "age": 72,

    "sex": "female",

    "fever": true,

    "cough": true,

    "shortness_of_breath": false

  }

  Response example:

  {

    "risk_level": "high",

    "id": 42,

    "created_at": "2025-08-14T00:00:00Z"

  }

- POST http://localhost:8000/predict (legacy — kept for tests/backward compatibility)

- GET  http://localhost:8000/predictions

- POST http://localhost:8000/train (provide records to train and activate best model)

- GET  http://localhost:8000/models (list trained models, active flag, and metrics)

OpenAPI docs are available at http://localhost:8000/docs and http://localhost:8000/redoc.

## Docker

- Build: docker build -t sekouai:latest .

- Run (Windows PowerShell):

  docker run --rm -p 8000:8000 -e SEKOU_SQLITE_PATH=/app/app.db -v ${PWD}\app.db:/app/app.db sekouai:latest

## Configuration

- SEKOU_SQLITE_PATH: path to SQLite file (default: app.db)

- SEKOU_CORS_ORIGINS: comma-separated list of allowed origins (default: *)

- SEKOU_LOG_LEVEL: logging level (e.g., INFO, DEBUG)

## Security & privacy

- CORS defaults to * for development; configure SEKOU_CORS_ORIGINS for production (e.g., https://yourclinic.example).

- Basic HTTP security headers are set (X-Content-Type-Options, X-Frame-Options, Referrer-Policy, X-XSS-Protection).

- No PHI/PII should be sent; this is an MVP with a simple rule-based model, not a certified medical device.

- Use a managed, encrypted database and audit logging for real deployments.

## Documentation

Architecture and design documents are in the `docs/` folder:

- Use case diagram: docs/use_case_diagram.png

- Sequence diagram: docs/sequence_diagram.png

- Class diagram: docs/class_diagram.png

- Deployment diagram: docs/deployment_diagram.png

- Architecture_UML_SekouAI.docx

- Technologies_MVP_SekouAI.docx

## Tests

- Run: pytest

- Tests cover health check, legacy prediction, and listing predictions.

## Project layout

- backend/

  - main.py (FastAPI app)

  - model_utils.py (triage + legacy rule-based models)

  - database.py (SQLAlchemy models and session)

  - schemas.py (Pydantic models)

- frontend/

  - index.html (UI), static/style.css (styles)

- requirements.txt

- Dockerfile