A high-fidelity, production-grade disease monitoring and risk prediction platform.
This system leverages Machine Learning to analyze historical outbreak data, weather patterns, and real-time reports to provide actionable insights for public health officials and the general public.
Unlike traditional systems, it focuses on detecting early weak signals — identifying outbreaks before they become obvious.
| Dashboard | Risk Map |
|---|---|
 |
 |
| Analytics |
|---|
 |
Most current outbreak detection systems:
- Reactive, not Proactive: Depend on confirmed hospital data which is often delayed.
- Late Detection: Outbreaks are identified only after significant community spread.
- Data Silos: Lack of integration between environmental factors and clinical data.
👉 This platform bridges the gap by detecting early signals before they escalate into crises.
Instead of waiting for large spikes, the system analyzes "Silent Signals":
- Micro-Trends: Gradual increases in symptoms or cases over 7–14 days.
- Environmental Context: Real-time integration of humidity, rainfall, and temperature data.
- Pattern Recognition: Regional disease patterns compared against historical baselines.
- AI-powered predictions using Random Forest and Gradient Boosting models.
- Dynamic risk scoring (Low / Medium / High) updated as new data flows in.
- Explainable risk factors (Weather + Growth + Volume).
- Interactive maps with outbreak clusters and hyperlocal risk detection.
- Seamless navigation between global, regional, and area-specific data.
- Time-series tracking for multiple pathogens (Dengue, Malaria, Flu, etc.).
- Comparative analysis of current trends against historical seasonal averages.
- Secure CSV / Excel upload support for rapid data ingestion.
- Automated cleaning and normalization of clinical reports.
"Detecting the invisible before it becomes inevitable."
- Framework: React Native (Expo)
- UI/UX: Premium Glassmorphic Design System
- Visualization:
react-native-chart-kit,react-native-svg - Mapping:
react-native-maps
- Framework: React.js (Vite)
- Charts:
Chart.js/Recharts - Styling: Tailwind CSS / Modern CSS Variables
- Runtime: Node.js / Express.js
- Database: MongoDB (Atlas)
- Architecture: RESTful API with automated geocoding services.
- Language: Python
- Library: Scikit-learn
- Models: Random Forest Regressor, Gradient Boosting
- Serving: Flask / FastAPI microservice
- Weather Data: OpenWeather / WeatherStack API
- Geocoding: OpenCage / Nominatim API
graph TD
A[Clinical Data + Reports] --> B[Processing & Cleaning]
C[Weather & Env Data] --> B
B --> D[Trend Analysis last 30 Days]
D --> E[Weak Signal Detection]
E --> F[ML Risk Engine]
F --> G[Insight Generation]
G --> H[Visualization & Alerts]
├── Backend/ # Express server, API controllers, and DB services
├── mobile-app/ # React Native source code (iOS/Android)
├── ml/ # Python ML models and prediction engine
├── frontend/ # React.js web dashboard
└── screenshots/ # Visual documentation of the implementationcd Backend
npm install
npm startcd mobile-app
npm install
npx expo startcd ml
pip install -r requirements.txt
python server.py-
**Tejas Kulkarni-- -CSV Upload & Validation -Data Preprocessing & Cleaning -Input Handling & Error Checking -UI for Data Handling - Deployment and database handling
-
**Yashashri Rajput-- - Environment Data Integration - API Development & Integration - UI for Data Visualization
-
**Samruddhi Patil-- - Risk Calculation Logic - Outbreak Prediction & Scoring - UI for Analytics & Insights
- Real-time Integration: Direct hooks into hospital management systems.
- Push Alerts: Geofenced notifications for high-risk zones.
- Deep Learning: Implementing LSTM models for better time-series forecasting.
- Multi-Agent AI: Specialized agents for pandemic simulation.
Developed as part of the AI-Based Disease Outbreak Prediction System Project.