SaveLife

Project Story: Savelife

Inspiration: The 9-Day Delay

In June 2026, when conflict erupted in Jebel Moya (Sennar State), over 136,000 people fled their homes in a matter of days. By the time international aid groups arrived to assess the situation on July 5th, families had been living in bus stations and schools for over a week without steady food or water. We realized that in South Sudan and Sudan, humanitarian data is historical, but displacement is happening in real-time. We built Savelife to close this "9-day delay" and provide responders with the lead time needed to save lives.

How We Built It: The Intelligence Pipeline

We didn't just build an app; we built a three-stage "Early Warning Pipeline" that translates conflict into action:

Ingestion (The Pulse): Our system acts as a 24/7 listener to the ACLED API. The second a "war event" is reported—like the SAF vs. RSF clashes—the pipeline triggers.

Transformation (The Sieve): Raw conflict data is messy and full of text "noise." We engineered a custom Data Translation Layer to strip away the junk and extract the "Predictors": coordinates, event intensity, and historical movement vectors.

Projection (The Forecast): This clean data is fed into a predictive model. Using the Sennar case as a blueprint, the system recognizes that Gedaref State is the logical destination based on road access and previous displacement patterns from Khartoum.

The Case Study: Turning Data into Lead Time

To validate Savelife, we applied it to the Sennar-Gedaref corridor:

The Reality: Fighting began June 26. Formal assessments happened July 5. That is 9 days of responders being "blind."

The Savelife Intervention: The moment the fighting was detected, Savelife would have calculated Gedaref as the primary destination.

The Result: Local responders in Gedaref would receive an automated SMS on June 26, allowing them to set up water points and family reunification desks before the 136,000 people arrived.

Challenges: The "Multi-Stage" Problem

The biggest challenge was accounting for the fact that many people in this region are "serial displaced"—some have fled Khartoum, then Al Jazira, and now Sennar. We had to refine our Data Mapping (Serialization) to ensure our model understands these complex, multi-stage journeys so it can predict not just the next town, but the specific reception centers where aid is most needed.

What We Learned: Translation is Life

We learned that technology is most powerful when it is simple at the point of impact. We take complex, high-level conflict data and translate it into a simple SMS message. We remove the "data science" burden from local NGOs so they can focus on their actual mission: providing food, shelter, and reuniting lost children with their families.

Built With

• acled
• iom-data
• javascript
• postgresql
• python
• sklearn