Project SENTINEL (Safety Environmental Network for Threat Identification, Notification, and Emergency Logistics) is an AI-driven disaster prediction and alerting platform designed to enhance public safety and disaster response. SENTINEL integrates machine learning models with community-verified data to predict and respond proactively to environmental threats such as wildfires and floods. The platform provides valuable, real-time insights to authorities and residents in high-risk areas, enabling them to take swift action to mitigate potential disasters.
The core components of SENTINEL are:
- Wildfire Prediction Model: Predicts wildfire occurrence and severity in California using historical wildfire data and climate information.
- Flood Prediction Model: A forthcoming model to predict flood risk based on weather, geographic, and community data.
- Community Engagement and Alert System: Integrates public input to enhance model accuracy and facilitate efficient emergency response.
"Predict. Notify. Protect—Together."
Project SENTINEL provides real-time disaster alerts with 88% accuracy, aiming for 92-95% by integrating additional data sources. Verified community alerts empower local authorities and residents, establishing SENTINEL as a proactive solution for reliable disaster resilience and preparedness.
The goal of Project SENTINEL is to harness the power of AI to predict, alert, and prevent environmental disasters. The key objectives are:
- Predict Wildfires and Floods: Use predictive analytics and machine learning to anticipate when and where environmental disasters will occur.
- Empower Authorities: Provide local governments and emergency response agencies with real-time alerts and insights to improve disaster preparedness and resource allocation.
- Engage Communities: Incorporate community-verified data into the prediction models to enhance accuracy, encourage public involvement, and improve response effectiveness.
- AI Prediction Models: Predicts wildfires and floods using advanced AI with high accuracy.
- Automated SOS Alerts: Collects and prioritizes emergency signals from social media.
- Community Engagement: Gamified reward system encourages users to report accurate incidents, improving the data quality and fostering community involvement.
- Data Fusion: Unique blend of AI-driven predictions and community input for hyper-local accuracy.
- Scalability: Adaptable across regions and disaster types, making SENTINEL a comprehensive global solution.
- Wildfire and Climate Data:
- Collected from public sources, such as government databases and weather stations.
- Preprocessed to remove irrelevant or incomplete records, and mapped wildfire occurrences to corresponding weather data for a unified dataset.
- Feature Engineering:
- Extracted relevant features, including temperature, precipitation, and fire size.
- Used one-hot encoding for categorical data (e.g., locations) and cyclical encoding for temporal features (e.g., month).
- Classification Task: Uses a Gradient Boosting Classifier to predict whether a wildfire will occur in a specific location and time.
- Regression Task: Predicts the area affected by wildfires using a Gradient Boosting Regressor (currently under development).
- Model Explainability: Employed SHAP to interpret feature importance and provide transparency regarding the key factors influencing predictions.
- Will leverage historical flood data and climate conditions to predict flood risk and impact.
- Community Alerts: Integrates community-reported incidents through social media and other platforms to validate and supplement model predictions.
- Gamification: A badge reward system will encourage community members to contribute data, ensuring sustained engagement.
"SENTINEL: The Waze for Disaster Alerts" — A comprehensive, AI-powered, community-driven platform for real-time wildfire and flood alerts, revolutionizing preparedness and emergency response for both authorities and local communities.
- Primary: Emergency agencies, municipalities, and fire departments needing precise real-time data to optimize response times and resource allocation.
- Secondary: Residents in high-risk areas, insurance companies looking for accurate risk assessments, and NGOs focused on climate resilience and advocacy.
- Prediction Accuracy: Improve model accuracy to achieve 92-95%.
- Response Time Reduction: Target an impressive 40% reduction in response times.
- User Growth: Aim to reach 25,000 active users within 18 months.
- Data Validation: Increase verified incident reports by an estimated 30%.
- Mobile & Web App: Real-time alerts and reporting for residents.
- Dashboard for Authorities: Centralized data for emergency response.
- Partnerships: Collaborations with government agencies and NGOs.
- Premium Subscription Plans: For alerts and in-depth insights targeted to residents in high-risk zones.
- Risk Data Licensing: Insights tailored and sold to insurance companies for effective risk management.
- Government Solutions: Customizable dashboards and analytics for government and municipal agencies, supporting disaster preparedness and emergency response.
Current Status: The Wildfire Prediction Model is functional, and work on the Flood Prediction Model is ongoing. Community engagement features are being prototyped for integration.
Project SENTINEL is an evolving platform, and we aim to:
- Complete the Flood Prediction Model to provide similar insights for flood risks.
- Expand to Other Regions beyond California to cover other high-risk areas.
- Integrate Real-Time Data Sources, such as satellite imagery, to improve the accuracy and timeliness of predictions.
- Community Dashboard: Create a user-friendly interface to display real-time alerts and allow community members to contribute data seamlessly.
- For Residents: Early warnings, real-time disaster alerts, and access to preparedness tips.
- For Authorities: Improved disaster management with centralized alerts and accurate predictions.
- For Insurance Companies: Risk data insights for effective risk assessments and planning.
Project SENTINEL is licensed under the MIT License. See LICENSE for more details.
Note: This repository is shared for evaluation purposes only as part of the Microsoft Founders Hub submission. All rights to the code, models, and related documentation are reserved by the author. Unauthorized use, modification, or distribution is strictly prohibited.
- CAL FIRE and CIMIS for providing the historical wildfire and climate data.
- Open-source community for contributing useful tools and frameworks.
- Local Community Members whose data contributions are instrumental in improving the platform's accuracy.
Project SENTINEL aims to create a safer, more informed world by leveraging AI for disaster prediction and emergency response. Let us know if you want to get involved or learn more about our mission to empower communities in times of crisis.