When an individual experiences a traumatic injury that requires emergency surgery, sometimes choosing not to operate can be what saves the patient’s life. In Dallas, those decisions are now being guided by artificial intelligence that’s leading to a technological shift in the industry.
Here is how it often plays out: A patient arrives with severe injuries from something like a car accident or a significant fall. Trauma surgeons are trained to move quickly and fix problems but aren’t the only providers who make decisions; the intensive care unit team is another stakeholder. The surgeons and ICU physicians can, at times, disagree about whether a patient is stable enough to undergo surgery, and each gives weight to opposing variables to make their cases.
“The physicians were literally verbally arguing on the floor to say when to take patients to surgery,” says Albert Karam, vice president of data strategy and analytics at Parkland Center for Clinical Innovation, or PCCI. “There wasn’t a standard in the industry for quantifying what the decision should be.”
Until now. As AI becomes more mainstream, the conservative medical community is allowing AI to impact decision-making in ways that would have been unthinkable a few years ago. As is often the case in the modern economy, AI and cloud technology collect reams of data that can improve outcomes and reduce costs, but it isn’t always used. Physical trauma is the leading cause of death for those under 45 and costs the economy $670 billion each year. With extensive electronic health records holding dozens of data points on any patient that passes through the system, the information care teams need is already there. However, these data banks need AI to make information accessible and display it in a way that can quickly guide decision-making.
At Parkland Health, an AI collaboration between Dallas County’s safety net hospital and PCCI is now providing ICU and trauma surgery teams with a data-based method to guide choices for treating traumatically injured patients. It is changing the paradigm for trauma surgery, where for years best practice has been to do surgery as early as possible to improve outcomes. But AI is providing a more precise outlook.
Parkland is the ideal candidate for a timesaving and outcome-improving technology. It is consistently the busiest emergency department in the country and was the only ED in the U.S. to see more than 200,000 patients in 2022 when it cared for 226,178 patients.
The Parkland Trauma Index of Mortality is a groundbreaking program that provides care teams with objective data about whether the patient is strong enough to undergo surgery, replacing what had been subjective decisions made in stressful moments.
The index analyzes dozens of variables and compares them to historical outcomes to give the patient a score determining whether they are ready for surgery. The score tells the providers if a patient is likely to die in the next 48 hours and, thus, requires stabilization and provides updates based on the patient’s evolving condition. Using a scale from 0 (great condition) to 100 (won’t survive surgery), physicians are able to make a data-driven decision about whether a patient is ready for surgery. After the first 12 hours, the score is updated hourly to determine whether a patient is likely to survive for the next 48 hours. Not only does it provide objective data to guide decisions, it also saves time. Each of the dozens of variables can take up to seven clicks by a physician to find. In the end, the call is still in the hands of the physicians, but now they have thousands of patient encounters and dozens of variables to inform them. The program, which went live in 2019, has revolutionized the trauma surgery process at the hospital.
“You have to look at multiple variables. A really smart human being can consider two or maybe three things at the same time in their head, but you can’t look at 50,” says Dr. Adam Starr, who championed the program at Parkland. “But the computer can.”
A professor in orthopaedic trauma at UT Southwestern Medical Center and the medical director for orthopaedic surgery at Parkland Hospital, Starr remembers the first patient whose score told the physicians he wasn’t ready for surgery. It was a young man in his 30s who had been in a motorcycle accident and had a femur fracture. He was awake and talking but had multiple broken ribs. When they received his PTIM score, it was much higher than expected, meaning he might have died on the operating table. If it were up to Starr’s previous training and prevailing dogma, they would have gone ahead with surgery, and according to PTIM, he likely wouldn’t have made it. They waited until his score went down to operate, and everything turned out well.
“I cannot imagine going back to what we had before because, in the past, we made our decisions based on our gut,” Starr says. “But that’s certainly not scientific, and it’s often wrong. This is a much better way to do it.”
Author
