An Automated Multi-Source Disaster SOS Verification and Intelligent Emergency Response System

Abstract

One of the most persistent problems in emergency management is the time gap between the moment a disaster begins and the moment a responding authority has enough verified information to act. Apex Response is designed to reduce this gap by allowing individuals affected by earthquakes, floods, or landslides to submit SOS reports with automatically captured GPS location and optional text, voice, photographs, and video. A multimodal language model, Gemini, extracts relevant information for triage, including the reported location, affected or injured people, possible entrapment, and other situational cues. The extracted information is treated as supporting evidence rather than as an independent decision. Each report is processed through an automated verification pipeline that checks appropriate external evidence sources, including government seismic records for earthquakes, weather-service data for floods and landslides, and regional news coverage. When external sources do not provide confirmation, the system checks for nearby SOS reports of the same disaster type received within five minutes. Based on the available evidence, reports are classified as Verified, Suspicious, Manual Verification, or Fake. Reports requiring human intervention can be reviewed through an authority dashboard. Once an emergency is verified, the system recommends a suitable rescue team and calculates a route using Dijkstra's shortest-path algorithm. The routing cost considers estimated travel time, traffic delay, blocked roads, flood or landslide damage, and damaged bridges. The platform is implemented as a Progressive Web Application using React, Tailwind CSS, Node.js, Express.js, Supabase, Leaflet, OpenStreetMap, Gemini API, Service Workers, and IndexedDB. Testing covers primary and secondary evidence matching, multi-report correlation, multimodal evidence handling, and rescue-route computation. The paper presents the system architecture, verification workflow, routing approach, implementation, testing, limitations, and future directions. Future work includes peer-to-peer communication between devices, richer sensor and AI integration, multi-team routing, and integration with government disaster-management systems.