TECHNOLOGIES DEVELOPED FOR THE FIELD ARE CHANGING THE LANDSCAPE OF EMERGENCY MEDICINE

NIKI KAPSAMBELIS

In the early morning darkness of Jan.18, 2018, Staff Sgt. Luke Scuilli and a fellow Green Beret entered a building in southern Afghanistan, planning an attack with a small group of commandos.

In the next moment, the building collapsed around them, changing Scuilli’s life forever. Unbeknownst to the soldiers, the building had beenbooby trapped with explosives, which detonated when they entered.

“I just remember a really loud explosion going off, seeing some rubble and debris,” said Scuilli, who served as the unit’s assault team leader and senior medical sergeant. Briefly losing consciousness, he woke a few second later to a pitch-black world; his night vision goggles had been knocked off by the explosion.

He also couldn’t move. At first he thought he’d been pinned under rubble; in reality, he had suffered a severe spinal injury.

The explosion represented the first domino to fall in a chain of events that brought Scuilli back to his hometown of Pittsburgh and more specifically to Carnegie Mellon’s School of Computer Science. He is now a senior research analyst for a team working to build robots driven by artificial intelligence that can amplify what Scuilli did as a medic in order to save more people like Scuilli the casualty.

“One of the biggest things [the explosion] opens up is: I’m obviously a certified and qualified medic for my team, but that doesn’t mean the medic doesn’t get hurt,” explained Scuilli. “Somebody has to step up and take care of me.”

And that somebody just may be a robot in a backpack, thanks to a project led by Artur Dubrawski, a research professor at the Robotics Institute.

FROM BATTLEFIELD TO HOMELAND

Dubrawski serves as the principal investigator of RoboTrac (Robotic Trauma Care in the Field), a project that also involves Robotics Institute faculty Howie Choset, Chris Atkeson and John Galeotti.

This Department of Defense-funded endeavor seeks to develop robotic and AI solutions for the field by multiplying the capabilities of medics. It requires pushing boundaries of what can be accomplished in practice by using advanced ultrasound imaging, precise robotic manipulation, soft and compliant actuation, and artificial intelligence to automate complex tasks in austere settings that are currently performed by highly-trained medical personnel.

Ultimately, the technology could help those who are injured in remote areas, far from any hospital or other emergency medical care, or in traffic accident or natural disaster scenarios where immediate access to advanced intensive care may not be readily available.

“It looked pretty obvious that many young men and women died after being wounded in combat situations, but they could be saved if sufficient support had been provided to them,” Dubrawski said.

He works on the project in partnership with his close friend, Dr. Ronald Poropatich, a professor of medicine at the University of Pittsburgh and director of Pitt’s Center for Military Medicine Research, Health Sciences. Dubrawski also has worked with critical care physicians and nurses at Pitt’s medical school for more than a decade.

Poropatich, deployed several times while serving as a pulmonary critical medicine physician in the U.S. Army, leads a sister project called TRAuma Care In a Rucksack (TRACIR), which Dubrawski also supports. Both projects include Pitt critical care medicine faculty Dr. Michael Pinsky and emergency medicine Drs. Frank Guyette, Lenny Weiss and David Salcido, reflecting a true team science approach. Both TRACIR and RoboTrac, which were funded by the U.S. Department of Defense in 2019, focus on military and civilian applications. The projects draw on large hospital data sets provided by medical helicopters flying around four states as part of the UPMC network, as well as data from all stages of intensive care from surgeries to step-down and intensive care units.

“The amount and complexity of this data are beyond human capabilities to comprehend and internalize in real time, so it’s an excellent target for AI,” explained Dubrawski. The tools currently in development quickly prioritize information for health care professionals, improving the speed and accuracy of the decisions to create better outcomes at a lower cost.

According to Poropatich, interest in the concept of AI medical support in the battlefield picked up after an October 2017 ambush of the Army’s 3rd Special Forces Group in Niger, killing four Green Berets. A drone flying overhead provided reconnaissance, but had no medical capability.

“That was a wakeup call to the Army,” said Poropatich.

The project develops robotics that must be small enough to fit into a backpack that a drone could drop near the wounded or injured person. The backpack would then open on its own, and a vest would inflate and crawl to the patient and mechanically stabilize them. The vest would be embedded with ultrasound and vital sign sensing capabilities that diagnose what’s wrong and inform necessary treatment.

In hospitals, equipment in the intensive care unit collects much needed diagnostic information. But according to Dubrawski, these bedside monitors often raise false alarms, saturating the attention of clinical personnel. His group has been working with clinicians at Pitt to develop a highly precise algorithm that fishes out the real alerts. The team is testing the tool in a pilot study. Poropatich said the team used data from approximately 60,000 trauma patients ranging in age from 18 to 55 and mapped the data to their medical records, with identities removed, to determine what treatments led to better outcomes.

The information allows the team to tell its robots what to do to optimize emergency care in the field. Using the same technology, the sensors in the robotic backpack would accurately measure what the wounded soldier needs, then direct the robots to perform life-saving functions, such as injecting fluid into the veins or inserting a tiny balloon via a catheter into an artery to stem the rapid loss of blood. Work also includes robots that can treat a collapsed lung or a blocked airway. And then ideally, according to Poropatich, the drone could take the wounded person off the battlefield.

To date, the team has created the early stages of some individual components and is testing the capability of robotic needle insertion as well as ultrasound-guided visualizations into models of human tissue built to resemble a body, known as human phantoms.

But to truly recreate what a medic does, they needed somebody to mimic. And that somebody, as it turns out, is Luke Scuilli.

THE LONG ROAD HOME

After the explosion in Afghanistan, Scuilli was first taken to Germany for spinal decompression surgery before undergoing reconstructive pelvic surgery at Walter Reed National Military Medical Center in Bethesda, Maryland. He also underwent spinal surgery and spent five months using a wheelchair while completing extensive physical and occupational therapy.

Among his many visitors was former Pittsburgh Steelers offensive tackle Alejandro Villanueva (TPR 2009), who won a Bronze Star as an Army ranger. When Villanueva visited, he brought Ron Poropatich with him. And Poropatich told Scuilli about the project.

At the time, Scuilli needed to concentrate on his recovery, which included learning how to walk again. But a year later, after his discharge, he moved back home to the Pittsburgh area, reached out to Poropatich and asked to hear more.

“I was trying to think of ways that I could do something fulfilling and something that would utilize my expertise without moving to the D.C. area,” explained Scuilli, who had been a firefighter for the city of Beaver Falls, Pennsylvania before joining the Army. So Poropatich put him in touch with Dubrawski.

“CMU has the greatest robotics team in the world, and the University of Pittsburgh has some of the greatest physicians in the world,” said Scuilli.

“What they didn’t have was somebody to consult with on battlefield medicine.” Scuilli sees his role as advising the team on what protocols most closely resemble a medic’s work in a combat setting. For example, the anesthesia a physician uses in an operating room differs from what a medic would carry into the field.

Ultimately, the project — which the Army would like to see completed by 2028 — could prove to be a game changer for civilian patients, say Dubrawski and Poropatich, particularly in rural parts of the country where trauma care is scant or nonexistent, and where hospitals are rapidly disappearing. “The concept [focuses on] how we augment care with technology, with clinical care support tools, with algorithms that are driven either autonomously or semi-autonomously,” Poropatich said.

For Scuilli, working on the project has been both personally and professionally rewarding. “It wasn’t until I got involved in this project that it showed me there as more to life than being a Green Beret. At first, I didn’t see it. It wasn’t until I found something to dedicate my focus and attention to that it really started turning things around for me,” Scuilli said. ■

EDITOR’S NOTE: While fulfilling his senior researcher duties at CMU, Luke Sciulli volunteered his hands-on medic expertise in a COVID-19 intensive care hospital in New York City, set up by his fellow former and current special operations soldiers. He attended to critically ill patients and served as a hospital supervisor.