Hyd-DSS — Emergency Decision Support

Overview

Hyd-DSS is an emergency decision-support ecosystem built to cut dispatcher overload and allocate first responders better. It replaces the existing first-come-first-served approach with a severity-based algorithm that sets priority dynamically.

Around that core sit AI hotspot detection, a human-in-the-loop dispatch engine, and a citizen-facing SOS portal.

The problem

In a first-come-first-served queue, a minor call that arrives first is handled before a critical one that arrives a minute later. When incidents spike, dispatchers are left triaging by hand, and spotting patterns across calls depends on someone noticing them.

How it works

Severity-based priority. Incidents are ranked by severity and re-prioritized dynamically, instead of by arrival order.
AI hotspot detection. The system continuously analyzes incident density, severity and proximity, clustering things like localized dengue cases or fire reports, and flags potential threats instead of relying on manual observation.
Human-in-the-loop dispatch. A recommendation engine suggests how to route ambulance, fire or police units, while final control always stays with the operator.
Citizen SOS portal. A public portal with auto-geolocation and live maps takes SOS reports.
A conversational front door. In the demo, an AI emergency assistant takes the report in plain language, flags its priority (a high-priority fire, say), gives immediate safety steps and confirms the unit that’s been dispatched.

Key features

01
Severity triage

Priority comes from how serious an incident is, not when it arrived.
02
Hotspot detection

AI clustering of incident density, severity and proximity.
03
Dispatch recommendations

Suggested routing for ambulance, fire and police units.
04
Operator in control

Recommendations only. A human makes every final call.
05
SOS portal

Auto-geolocated distress reports from citizens.

Challenges & what I learned

The hard part of a system like this is trust: a severity score or a hotspot flag is only useful if operators believe it. Keeping a human in the loop for every dispatch is what makes the automation acceptable in the first place.