VitalWatch: Real-Time NCD Monitoring Platform

Download the VitalWatch Presentation

One-Line Summary: A real-time IoT platform that ingests wearable glucose and blood pressure streams into GridDB Cloud, enabling trend detection, predictive spike alerts, and doctor notifications via SMS designed for 1 lakh+ patients in underserved rural clinics.

---

🚨 The Problem: India’s Silent Health Emergency

The Story of Ramaiah: In a small primary health center in rural Karnataka, Dr. Meera sees 80 patients daily. She has exactly 4 minutes per patient. What she cannot see is that Ramaiah, a 54-year-old farmer, has had his fasting glucose creeping from 140 to 190 over the past three weeks. She cannot see that his blood pressure spikes to 165/105 every night around 2 AM.

Two weeks later, Ramaiah is rushed to the district hospital with diabetic ketoacidosis. This story repeats millions of times across India every year.

Why Current Approaches Fail:

• Single readings are misleading: A single fasting glucose reading tells you almost nothing about the other 23 hours.
• Data is lost: Readings are written on paper slips or isolated devices that never connect to anything.
• Doctors are overwhelmed: A rural physician managing 1,000+ chronic patients cannot possibly track each one individually without a system to flag urgent cases.

---

💡 Our Solution: VitalWatch

VitalWatch is built on one core belief: chronic disease management should be continuous, predictive, and proactive—not episodic, reactive, and crisis-driven.

How It Works

1. Data Collection: Patients wear low-cost continuous glucose monitors and blood pressure wristbands.
2. Transmission: Devices transmit readings via Bluetooth to a local gateway (Android phone/Hub), which publishes to an MQTT broker.
3. Ingestion: A Node.js service validates and writes data into GridDB Cloud in real-time.
4. Analytics: Our engine calculates rolling averages, detects anomalies, and uses predictive models to estimate spike probabilities.
5. Alerting: High-risk events (e.g., glucose > 300) trigger SMS alerts to doctors via Twilio and WhatsApp messages to families.
6. Visualization: Doctors access a Grafana dashboard to view patient trends and population health metrics.

---

🛠️ GridDB Cloud Integration

GridDB is the foundation that makes VitalWatch possible.

Why GridDB?

Chronic disease monitoring is a time-series problem. We need to answer questions like:

• “What was this patient’s average glucose over the past 7 days?”
• “Show me all readings above 200 in the past 24 hours.”

GridDB was chosen because it handles millions of time-series readings without performance degradation, unlike traditional relational databases.

Container Design

• Glucose_readings (Time-Series): Stores timestamp, patient ID, glucose value (mg/dL), device ID, and meal context.
• BP_readings (Time-Series): Stores systolic/diastolic values, pulse rate, and body position.
• Alerts_log (Time-Series): Logs every system-generated alert, severity, and acknowledgement status.
• Patients (Collection): Master data including demographics and personalized thresholds.

---

💻 Tech Stack

Component	Technology
Database	GridDB Cloud (Time-series engine)
Backend	Node.js (Express), GridDB-node-api
IoT Connectivity	MQTT.js, Eclipse Mosquitto
Analytics/ML	Python 3.10, Pandas, Scikit-learn (Isolation Forest), TensorFlow (LSTM)
Alerting	Twilio SDK (SMS), Firebase
Visualization	Grafana
DevOps	Docker, GitHub Actions

---

📅 Implementation Plan

• Phase 1 (Week 1): Core Infrastructure. Set up GridDB Cloud, MQTT broker, and sensor simulator.
• Phase 2 (Week 2): Alerting System. Implement threshold alerts and Twilio integration.
• Phase 3 (Week 3): Visualization. Configure Grafana dashboards with live GridDB data.
• Phase 4 (Week 3-4): Analytics. Build Python service for rolling averages and LSTM predictive alerts.
• Phase 5 (Week 4): Polish & Documentation. Performance testing and demo recording.

---

🚀 Traction & Vision

• Pilot: Identifying 3 Primary Health Centers (PHCs) in Karnataka for a 500-patient pilot.
• Scale: Expand to 10,000 patients across 15 PHCs in months 4-8.
• Long-Term: Partner with state NCD programs to scale to 100,000+ patients.

Team Contact:

• Manthan V. Done (manthandone07@gmail.com)
• Omkar B. Kore (omkarkore80.90@gmail.com)