Upload folder using huggingface_hub

.gitignore

@@ -0,0 +1,11 @@
+__pycache__/
+*.pyc
+*.pkl
+*.db
+.env
+venv/
+.DS_Store
+simulator/offline_buffer.jsonl
+*.log
+.vscode/
+.idea/

README.md

@@ -0,0 +1,70 @@
+# 🏥 Smart Health Monitor
+
+Edge-AI health monitoring system with **offline-first architecture**, **multi-signal fusion**, and **adaptive personal baselines**.
+
+---
+
+## Quick Start (3 terminals)
+
+```bash
+# 1. Install dependencies
+pip install -r requirements.txt
+
+# 2. Train the model (run once)
+python -m ml.train_model
+
+# Terminal A — Start Flask API
+python -m backend.app
+
+# Terminal B — Run IoT simulator
+python -m simulator.simulate --patient_id 1 --count 60 --anomaly_rate 0.2
+
+# Terminal C — Launch Streamlit dashboard
+streamlit run dashboard/app.py
+```
+
+---
+
+## Architecture
+
+```
+IoT Sensor → Kalman Filter → Edge ML → Multi-Signal Fusion → Tiered Alert
+                                                ↓
+                                        SQLite (offline buffer)
+                                                ↓
+                                     Sync to cloud on reconnect
+```
+
+---
+
+## Alert Levels
+
+| Level | Trigger          | Action                          |
+|-------|------------------|---------------------------------|
+| 0     | No anomaly       | —                               |
+| 1     | Mild (score>0.4) | Local buzzer                    |
+| 2     | Moderate (≥2 signals) | SMS via GSM / 2G           |
+| 3     | Critical (score>0.85 or 3+ signals) | Push notification |
+
+---
+
+## API Endpoints
+
+| Method | Path                        | Description              |
+|--------|-----------------------------|--------------------------|
+| GET    | `/api/patients`             | List all patients        |
+| POST   | `/api/patient`              | Register new patient     |
+| POST   | `/api/vitals`               | Ingest sensor reading    |
+| GET    | `/api/vitals/<id>`          | Get vitals history       |
+| GET    | `/api/alerts/<id>`          | Get anomaly alerts only  |
+| GET    | `/api/baseline/<id>`        | Get adaptive baseline    |
+
+---
+
+## Key Innovations
+
+- **Adaptive Personal Baseline** — Welford online algorithm, no raw history stored
+- **Multi-Signal Fusion** — Alert requires ≥2 independent signals to fire
+- **Kalman Filter** — Real-time noise reduction before ML inference
+- **Offline-First** — SQLite buffer + auto-sync on reconnect
+- **ABDM-Ready** — `health_id` field on every patient for Ayushman Bharat integration