AI Mass Casualty Triage System TriageAI Deployed in UN Humanitarian Aid: Casualty Classification Speed Increased 12x

AI Triage Officers at Disaster Sites

On January 10, 2029, the United Nations Office for the Coordination of Humanitarian Affairs (OCHA) announced the official deployment of the TriageAI mass casualty triage system in three refugee camps in the Horn of Africa. Developed by Swiss medical AI company MedTriage, the system can complete casualty triage at a speed of 60 people per minute during mass casualty events — 12 times faster than manual triage.

"In disaster sites, triage speed directly determines life and death," said OCHA emergency medical advisor Pierre Dupont. "Traditional START triage requires trained medical personnel to individually assess each casualty, taking about 60 seconds per person. TriageAI reduces this time to 5 seconds."

System Components

TriageAI consists of three components. The first is the portable vital sign monitoring armband VitalBand, which collects heart rate, blood oxygen, respiratory rate, and peripheral perfusion index in real time through optical sensors and bioimpedance analysis. Each armband costs only $12 and is reusable.

The second is the computer vision module TriageEye, a depth camera array installed on tripods that rapidly assesses injury severity by analyzing casualties' gaits, postures, and facial features. The system can identify critically injured casualties requiring immediate attention within 3 seconds.

The third is the central decision engine TriageBrain, which fuses VitalBand and TriageEye data and classifies casualties into four categories according to international START triage standards: red (immediate), yellow (delayed), green (minor), and black (unsalvageable).

First Real-World Deployment

TriageAI was tested in its first real mass casualty event following deployment. On January 13, a market explosion in Mogadishu, Somalia, injured approximately 180 people. Local medical teams used TriageAI to complete triage of all casualties within 15 minutes — traditional methods would have required approximately 3 hours.

The system classified the 180 casualties as 32 red, 67 yellow, 74 green, and 7 black. Subsequent verification showed triage accuracy of 94.2% and 96.8% consistency with experienced emergency physician judgments.

Limitations and Controversy

TriageAI's limitation lies in its inability to handle injuries requiring complex clinical judgment. For polytrauma, burn area assessment, and internal bleeding scenarios requiring clinical experience, the system still relies on human judgment.

Additionally, automated black tag (unsalvageable) assignment has raised ethical concerns. Critics argue that algorithms determining who isn't worth saving poses moral risks. MedTriage responds that black tags serve only as reference suggestions, with final decisions always resting with human medical personnel.