Feds at Work: Improving survival rates for hemorrhaging patients in transit
Developed an artificial-intelligence system enabling medics to detect bleeding in injured patients
When critically injured soldiers and civilians are hemorrhaging on the way to the hospital, their lives may hang in the balance. Now, a portable computer system can detect internal bleeding quickly and accurately during emergency transport. With that knowledge, medics can treat patients appropriately and trauma centers can get ready to provide immediate blood transfusions to save these patients’ lives.
Jaques Reifman, a senior research scientist with the Department of the Army, led the team that created APPRAISE — a first of its kind system that uses artificial intelligence and automated, pattern-recognition algorithms to interpret vital signs.
“The APPRAISE system addresses the number one killer on the battlefield by identifying casualties at risk for uncontrolled bleeding who must be immediately evacuated to a hospital,” said Anders Wallqvist, deputy director of the Army’s Biotechnology High Performance Computing Software Applications Institute.
The system can also help with civilian trauma cases, especially in remote areas where specialized trauma centers could be hours away.
For decades, the U.S. military has sought to develop technologies to identify casualties with internal bleeding. Reifman and his Army team, together with a group from Massachusetts General Hospital, have successfully tested the APPRAISE system over several years.
“Jaques Reifman saw the need to predict when someone is going to require a blood transfusion after watching 12 years of war.” ~ George Ludwig, U.S. Army Medical Research and Materiel Command
It has paved the way for Boston-area trauma units to be ready for immediate surgery and replenish lost blood without wasting time and resources on false alarms.
An APPRAISE unit initially was installed in two medical helicopters operated by Boston MedFlight. Between February 2010 and December 2012, the units collected vital sign data on 209 trauma patients transported to three trauma centers. The automated system successfully identified 75 to 80 percent of patients with life-threatening bleeding after monitoring them for just 10 minutes.
By 2015, APPRAISE had been tested on more than 1,000 civilian trauma patients.
“Jaques Reifman saw the need to predict when someone is going to require a blood transfusion after watching 12 years of war,” said George Ludwig, acting principal assistant for research and technology with the U.S. Army Medical Research and Materiel Command.
To create the system, Reifman assembled a team from many disciplines — engineers, software developers and medical clinicians. During their 10-year quest, group members faced numerous challenges.
The system had to operate successfully in uncontrollable environments. It needed to be fully automated so the data could be analyzed without human input or supervision — important for assisting combat medics who could be tired, distracted, inexperienced or overwhelmed by many casualties.
“Not too often do scientists have an idea or a dream and then over time advance the science to take the concept to deployment and show that it works.” ~ Jaques Reifman, U.S. Army Medical Research and Materiel Command
The system also needed to operate in real time, on mobile computing platforms with limited processing power and memory. And it had to work with missing and inconsistent sensor data because patient vital-sign information collected during transport is notoriously unreliable.
Key contributors to Reifman’s APPRAISE team included Maxim Khitrov and Jianbo Liu as well as Andrew Reisner, an investigator with the Army’s Biotechnology High Performance Computing Software Applications Institute and an emergency room doctor at Massachusetts General.
The Army received two U.S. patents for the system and is now pursuing Food and Drug Administration clearance and a licensing agreement with a commercial partner.
Reifman “argued for years that we measure many vital signs, but that we don’t know if the data we are collecting is actually information we need to be sure someone is in critical condition,” said Gary Gilbert, program manager for the Army’s Medical Intelligent Systems.
Reifman said the system offers “an opportunity to save lives on the battlefield” and eventually could assist emergency medical personnel throughout the country.
“Not too often do scientists have an idea or a dream and then over time advance the science to take the concept to deployment and show that it works,” he said. “I’m thankful that I’m here at the Army, which provides longevity in funding and support over time.”
Jaques Reifman and the APPRAISE Team are finalists for a 2016 Samuel J. Heyman Service to America Medal, or Sammies. Each year, the Partnership for Public Service honors federal employees whose remarkable accomplishments make our government and our nation stronger. For the second time, we will also present the annual “People’s Choice” award. Please vote for the person or team you find most inspiring. (Voting closes at 11:59 p.m. EST on September 9, 2016.)
