Hayat

Project Name

Hayat (حياة) — Real-Time Health Infrastructure Intelligence for Conflict Zones

Tagline

When hospitals go dark, Hayat shows you which ones are still standing — in real-time, in your language.

Track

Disaster Response

Inspiration

In 2023, the WHO documented over 2,000 attacks on healthcare facilities in conflict zones. In Gaza, 26 of 36 hospitals have been rendered non-functional. In Sudan, 70-80% of health facilities in conflict areas are destroyed. When hospitals are destroyed, people don't just lose buildings — they lose the ability to survive treatable injuries, deliver babies safely, manage chronic illness, and access life-saving medication.

Yet there is no real-time, publicly accessible system that tells a wounded civilian, a fleeing family, or a humanitarian aid worker which health facilities are still operational. The data exists — scattered across WHO spreadsheets updated days late, satellite archives that no one checks in real time, and situation reports written in languages the affected
populations can't read. People are dying in the gap between data and access.

We built Hayat because 120 million forcibly displaced people — the majority from Muslim-majority nations — deserve better than a spreadsheet updated last week.

What it does

Hayat (حياة — "Life" in Arabic) is a real-time health infrastructure intelligence platform that monitors the operational status of health facilities in active conflict zones. It fuses five humanitarian data sources into a single interactive
3D globe:

Conflict events from ACLED (Armed Conflict Location & Event Data) — real-time data on battles, shelling, and violence
against civilians with precise coordinates across 230+ countries
Health facility locations from WHO/HDX Humanitarian API — baseline data on hospitals and clinics across crisis regions
Satellite imagery from Copernicus/Sentinel Hub — before-and-after satellite photos showing structural damage to hospital buildings
Humanitarian reports from ReliefWeb — situation reports from WHO, OCHA, and MSF, summarized by AI (Google Gemini) to extract facility-level operational status
Displacement flows from UNHCR — refugee movement data showing where displaced populations are heading and in what numbers

Users can:

Explore an interactive 3D globe where every health facility in a conflict zone is color-coded by operational status: green (operational), yellow (damaged), red (destroyed)
Watch a time-lapse of how conflict erodes healthcare infrastructure over weeks and months — conflict events appear chronologically and hospital markers turn from green to red, one by one
Click any facility to see its satellite before/after imagery, AI-generated damage assessment, remaining capacity, and directions to the nearest operational alternative
View everything in Arabic, French, or English — covering the primary languages of the most affected populations in
Gaza, Sudan, Syria, and the Sahel
See displacement flow arcs showing where millions of refugees are moving — from Syria to Turkiye, from Sudan to Chad,
from Gaza to Egypt

How we built it

Architecture

Hayat uses a two-service architecture: a Next.js frontend for visualization and a Python FastAPI backend for data fusion and AI processing.

Frontend

Next.js 16 with TypeScript and Tailwind CSS for the application shell
deck.gl 9 with GlobeView for the 3D globe visualization — rendering conflict events as animated scatter points, health facilities as color-coded markers, and displacement flows as curved arc layers
DataFilterExtension for GPU-accelerated time-based filtering during the time-lapse animation
Custom satellite before/after comparison slider component
Responsive panel system for facility details and AI assessments

Backend

Python FastAPI handling all external API calls, data fusion, and caching
ACLED API for real-time conflict event data (battles, explosions, violence against civilians) with precise geo-coordinates
HDX HAPI (Humanitarian Data Exchange) for health sector operational presence data
ReliefWeb API for humanitarian situation reports from WHO, OCHA, MSF, and 4,000+ other sources
UNHCR Population API for refugee displacement flow data across origin and asylum countries
Proximity correlation engine that flags health facilities within radius of conflict events and computes damage probability scores

AI/ML Layer

Google Gemini 2.5 Flash for summarizing humanitarian reports and extracting facility-level operational status mentions
Sentinel Hub Process API for fetching before/after Sentinel-2 satellite imagery (10m resolution) of hospital buildings
DeepL API for translating all interface text and AI-generated assessments into Arabic and French

Data Fusion Logic

When a conflict event (battle, shelling, explosion) occurs near a health facility, Hayat correlates the event with the
facility using haversine distance. Cumulative nearby events generate a damage probability score. AI then cross-references this with ReliefWeb report mentions and satellite change detection to produce a confidence-rated operational status:
Operational, Damaged, or Destroyed.

Challenges we ran into

Correlating conflict events with health facilities at scale required optimizing the haversine distance calculation for thousands of events against dozens of facilities per animation frame
ACLED's OAuth authentication flow required careful token management and caching to avoid hitting rate limits during rapid development
Achieving smooth time-lapse animation with deck.gl's DataFilterExtension while simultaneously recomputing facility status colors required splitting GPU filtering (for conflict dots) from CPU computation (for facility status)
Satellite imagery from Sentinel Hub has a revisit period of 5 days — finding cloud-free before/after image pairs for specific hospital coordinates required experimenting with date ranges
Rendering Arabic text right-to-left while keeping the 3D globe interaction left-to-right required careful CSS isolation of the panel and overlay components

Accomplishments that we're proud of

The time-lapse animation of Gaza's hospitals going dark is genuinely haunting — watching 12 green markers turn red over 30 seconds makes the scale of healthcare destruction viscerally real in a way that statistics alone cannot
Fusing five separate humanitarian data sources (ACLED, WHO/HDX, UNHCR, ReliefWeb, Sentinel Hub) into a single coherent intelligence layer that generates facility-level assessments no single source provides
Full multilingual support in Arabic, French, and English — because the people who need this information most don't read English
The satellite before/after comparison for Al-Shifa Hospital — seeing the physical destruction of the largest hospital in Gaza from space is devastating and undeniable
Building a complete, deployed, functional prototype in 6 hours that could genuinely serve humanitarian organizations

What we learned

Humanitarian data is simultaneously abundant and fragmented — ACLED, UNHCR, WHO, and ReliefWeb all track overlapping aspects of the same crises but in completely different formats, with different update frequencies, and through different APIs
The gap between "data exists" and "data is accessible" is where people die — WHO tracks hospital attacks, but the information takes days to reach the field
deck.gl's GlobeView is extraordinarily powerful for visualizing geospatial crisis data — the 3D globe creates an emotional response that flat maps simply don't
AI summarization of humanitarian reports is viable and valuable — Gemini can extract facility-level status from dense WHO situation reports in seconds
Building for real crisis contexts forces design decisions that "move fast and break things" culture never considers — every UI choice carries weight when the user might be looking for the nearest hospital while wounded

What's next for Hayat

Immediate (1-3 months)

Expand coverage to all active conflict zones — Yemen, Myanmar, Ukraine, DRC, Ethiopia. The same data sources (ACLED,
Sentinel Hub, WHO) provide global coverage with no architecture changes.
Mobile-first version with offline capability for field workers in areas with limited connectivity — cache facility data locally and sync when connection is available.
SMS/WhatsApp integration — a civilian can text their location and receive the nearest operational health facility in their language. The 3D globe is for strategic planning; text alerts are for survival.

Who this helps

Displaced civilians (120+ million globally) who need to find operational health facilities after attacks — a mother in labor in Gaza, a child with shrapnel wounds in Sudan, a diabetic without insulin in a camp
Humanitarian aid workers (MSF, ICRC, WHO field teams) who need real-time operational awareness of health infrastructure to direct ambulances, position medical teams, and allocate scarce supplies
Humanitarian coordinators and planners making resource allocation decisions about where to deploy mobile health units, which facilities to prioritize for repair, and where to pre-position medical supplies
Journalists and advocacy organizations documenting the systematic destruction of healthcare in conflict zones — the time-lapse visualization is an undeniable, data-backed visual record
Policy makers and donors who need evidence-based intelligence about health infrastructure gaps to direct funding where it's most needed

Why it matters

Healthcare is protected under international humanitarian law. In reality, it's systematically targeted. The WHO Surveillance System for Attacks on Healthcare documented 2,000+ attacks in 2023 alone. In Gaza, 26 of 36 hospitals were
destroyed or severely damaged. In Sudan, 70-80% of health facilities in conflict areas are non-functional.

When a hospital is destroyed, the death toll extends far beyond the attack itself. Pregnant women can't deliver safely.
Trauma patients bleed out from treatable wounds. Diabetics lose access to insulin. Cancer patients miss chemotherapy. Newborns die from preventable infections.

The information about which facilities are still operational exists — but it's scattered across WHO spreadsheets,
satellite archives, and humanitarian reports in different languages, updated with days or weeks of delay. Hayat closes that gap by fusing these data sources into a real-time, multilingual, publicly accessible intelligence platform.

For the 120+ million forcibly displaced people in the world right now — the vast majority from Muslim-majority nations — this isn't a technology project. It's a lifeline. response.

Built With

nextjs typescript tailwindcss python fastapi deck-gl mapbox google-gemini sentinel-hub deepl acled unhcr reliefweb hdx-hapi vercel render