Painel responsável de vigilância epidemiológica sintética com sinais estatísticos interpretáveis.
Responsible synthetic epidemiological surveillance with interpretable statistical signals.
PT-BR · English · Live Demo · Stack · Architecture · Quick Start · Author
Responsible analytics notice: uses synthetic epidemiological-style data. Alerts and reliability scores are statistical decision-support, not clinical diagnosis, outbreak confirmation or SUS production systems.
O SentinelaSUS é um lab full-stack (FastAPI + Next.js) de vigilância sintética: overview, séries temporais, central de alertas, score de confiabilidade, qualidade de dados e briefing executivo.
Painéis epidemiológicos amadores misturam dado real, sem qualidade nem limite ético — e “alertas” sem método viram ruído ou medo.
Estudantes e profissionais de dados/saúde que querem treinar vigilância responsável com métodos interpretáveis em ambiente sintético.
- Overview de vigilância e KPIs
- Explorador de séries temporais
- Alert center com sinais interpretáveis
- Reliability / data-quality views
- Executive brief e comparação de métodos (conforme UI do lab)
- Pipeline de dados sintéticos + API Python
- Dados sintéticos — não é feed oficial do SUS
- Não diagnostica, não dispara resposta sanitária real
- Lab demo público ≠ sistema de vigilância em produção
SentinelaSUS is a FastAPI + Next.js lab for synthetic surveillance: overview, time series, alert center, reliability score, data quality and an executive brief.
Amateur epi dashboards mix real data without quality or ethical limits — and method-free “alerts” become noise or fear.
Data/health learners who want to practice responsible surveillance with interpretable methods on synthetic data.
- Surveillance overview and KPIs
- Time-series explorer
- Alert center with interpretable signals
- Reliability / data-quality views
- Executive brief and method comparison (as implemented in the lab UI)
- Synthetic data pipeline + Python API
- Synthetic data — not an official SUS feed
- Does not diagnose or trigger real public-health response
- Public lab ≠ production surveillance system
| Surface | URL |
|---|---|
| Public lab | https://sentinelasus.vercel.app |
| GitHub | see Repo badge above |
How to try: open the lab → review overview KPIs → explore a series → inspect alerts/reliability → read the executive brief.
 Home |
 Overview KPIs |
 Time series |
 Alert center |
 Data quality |
Hero |
| Layer | Technology |
|---|---|
| Web | Next.js, React, TypeScript, Tailwind, Recharts |
| API | FastAPI, Pandas, NumPy, Pydantic |
| Deploy | Vercel (vercel.json), optional Render (render.yaml) |
frontend/ Next.js UI
api/ | backend/ FastAPI services
data/ synthetic datasets
scripts/ generation / pipeline helpers
tests/ pytestFlow: synthetic events → aggregate → signal rules/stats → dashboard + brief.
Prerequisites: Node.js 20+, Python 3.10+.
# data + API (see repo scripts / run.py)
pip install -r requirements.txt
python run.py
# or: uvicorn on the api package as documented in-repo
cd frontend
npm install
npm run dev- Synthetic-first to avoid leaking real health microdata in a portfolio lab
- Interpretable signals over black-box “AI outbreak” claims
- Split UI/API so methods stay testable in Python
- More method cards and clearer uncertainty messaging
- Stronger data-quality drills
- Deeper demo scripts for interviews
Felipe Alirio Baruja — data / product / full-stack portfolio.
- Portfolio: https://barujafe.vercel.app/
- GitHub: https://github.com/BarujaFe1
- LinkedIn: https://www.linkedin.com/in/barujafe/
MIT — see LICENSE.