During the Napoleonic Wars, the French Imperial Army tried something rather novel: medics treated wounded soldiers according to the severity of their injuries instead of by military rank. Triage—prioritized care based on the severity of condition—has proved critical for saving lives on many battlefields since.

Now combat triage is undergoing yet another dramatic transformation—this time, a digital one. Working with the NATO HQ Supreme Allied Command Transformation Innovation Hub and the SACT Medical Branch, student engineers and scientists at Johns Hopkins University and the Czech Technical University have leveraged artificial intelligence to create a Digital Triage Assistant. Worn like a watch on soldiers’ wrists, the device collects their vital signs and location data, feeding that information in real time to a dashboard that not only tells medics where the wounded are located, but also assesses the severity of their injuries.

The student engineers are all members of the Clark Scholars Program. Established in 2016 with a $15 million investment in the Whiting School of Engineering from the A. James & Alice B. Clark Foundation, the program is designed not only to attract talented engineering students to Johns Hopkins, but also to prepare them for leadership roles. The cohort is advised by Lawrence Aronhime and Alexander Cocron, faculty members in the school’s Center for Leadership Education.

The team’s main innovation is a machine-learning model that combines data gathered from wearables to calculate a “mortality likelihood score” for an individual soldier, said Cocron. The score will help combat medics quickly determine who should receive treatment first.

During their research, the students discovered that current battlefield triage comes down to a few simple assessments: Can the soldier stand up? Can they raise their hand? Are they able to yell, or are they unconscious? With very little data, combat medics must evaluate injuries and prioritize evacuation in the shortest time possible.

“We were surprised to find that medics decide if someone is on the next helicopter out based on just two or three factors. That’s born out of necessity–they can’t afford to spend even a few minutes triaging a single solider. But we don’t have that limitation with a digital triage assistant. The system allows for continuous and real-time monitoring of vitals, and alerts the medic if a soldier’s condition gets worse,” said David Calvo, a fourth-year computer science major and team member.

The students designed the Digital Triage Assistant with an eye toward automating this process, getting the right information to medics as quickly as possible.

“We found we could model the triage process well in a programming language. But these calculations would be very hard for a human to do on the go,” said Jonathan Edwards, also a fourth-year computer science major and team member.

In addition to creating the mortality likelihood score, students built a map-like dashboard that uses multi-colored dots to indicate each soldier’s score and GPS location. The system also shares live triage updates with commanding officers at headquarters.

This article was excerpted from the Hub >>