Jannah Theme License is not validated, Go to the theme options page to validate the license, You need a single license for each domain name.
California NewsUSA

Can artificial intelligence make sepsis less deadly? UCSD Health study suggests yes

Sepsis is a life-threatening response to infection, which can cause runaway inflammation and a cascade of organ damage. It is estimated to kill at least 350,000 Americans a year.

Early detection of this disease is essential to saving lives. Medical research shows that the sooner medical teams can detect signs of sepsis, the sooner they can begin life-saving antibiotic treatments and administer intravenous fluids, supporting the body’s efforts to regain balance.

A team of researchers and clinicians at UC San Diego Health worked to see if artificial intelligence could provide an advantage in the early diagnosis of sepsis, and in a paper published Tuesday, the group finds that a system developed in internal called “DIAL,” a machine learning model trained with more than 100,000 digital records of sepsis patients can reduce mortality rates.

Every hour, COMPOSER reviews the electronic health records of UCSD emergency patients, examining myriad factors such as prescribed medications and most recently collected vital statistics to predict who based from its prior massive collection of health records, could be in the early stages of sepsis.

The real value of this learning algorithm, explained Dr. Gabriel Wardi, an emergency medicine and sepsis specialist and one of the paper’s co-authors, lies in the gray area between sepsis and many other conditions can mimic this condition.

“The majority of sepsis that occurs in the emergency department could probably be assigned to a medical student or an undergraduate student,” Wardi said. “They have a high fever, their heart beats very fast, their blood pressure is low; you don’t need an AI algorithm to say, “Hey, this person is septic.”

But some cases are not so obvious. Patients often arrive with unclear symptoms, such as general weakness, that could indicate early sepsis or many other medical problems.

“Often what happens is they have an initial workup, blood work will be done, maybe imaging studies, but (emergency departments) are busy places, it It’s difficult for providers who have done that initial assessment to say, ‘Hey, let me come back and put everything together,'” Wardi said. “That’s the advantage of the algorithm.

“This helps our providers in situations where there is some diagnostic uncertainty. It’s almost like a spider feeling, (saying) ‘hey, this person seems to be at high risk of developing sepsis in the next few hours, why don’t you go check them out again and see if they have any. would benefit.’ fluids, antibiotics, cultures to see what happens? »

Published in the journal npj Digital Medicine, the COMPOSER study involved a total of 6,217 emergency patients at UCSD’s two emergency departments in Hillcrest and La Jolla. Researchers compared the results of more than 5,000 emergency room patients seen between January 1, 2021 and December 6, 2022 to those of 1,152 emergency room patients treated while COMPOSER was active for almost five months, from December 7, 2022 to April. 30, 2023.

The researchers calculated a sepsis mortality rate of 9.5 percent for those supervised by COMPOSER, which is 1.9 percentage points lower than the predicted mortality rate of 11.39 percent. The group seen without active COMPOSER had an adjusted mortality rate of 10.3 percent.

The results, the article acknowledges, should be seen as correlations rather than proven causes and effects, because the trial was not randomized.

Dr. Karin Molander, director of the Sepsis Alliance, a nonprofit organization that works to reduce the prevalence of the disease nationwide, reviewed the paper and said that while more work remains In-depth, the results were encouraging.

While saying doctors will still want to be able to verify the facts on which AI advice is based, Molander discussed the idea of ​​having a “super smart” assistant with the time to sift through voluminous medical records and figure out trends, then compare. these trends to the experiences of thousands of other patients, is exciting.

“Knowing that there’s an AI in the background that doesn’t require sleep, bathroom breaks, or meals to help you monitor the system, that sounds pretty good,” Molander said. “But the challenge is to make sure he’s not hallucinating or, God forbid, fabricating.

“He cannot come to wrong conclusions.”

The researchers calculated that the 1.9 percent drop in mortality, although small, resulted in 22 patients surviving sepsis who otherwise would have died. The impact appears to have been most visible at UC San Diego Medical Center in Hillcrest, which reportedly “experienced a significant decrease in mortality” during the COMPOSER trial period, while Jacobs Medical Center in La Jolla “did not did not observe any significant change. .”

Obtaining observable results required significant work to tune the COMPOSER algorithm, allowing it to signal uncertainty and avoid sending too many false alarms to already busy caregivers. The algorithm alone cannot do anything. All actual care ordered for patients must come from licensed healthcare professionals.

Co-author Shamim Nemati, an associate professor of bioinformatics at UCSD and a doctorate in electrical engineering and computer science from the Massachusetts Institute of Technology, said the key was finding a way to analyze conditions that mimic sepsis but are actually something else.

“We had to teach the algorithm to distinguish between sepsis and lookalikes, you know, liver cirrhosis, liver failure and (gastrointestinal) hemorrhage,” Nemati said.

UCSD is already expanding COMPOSER’s real-time analysis beyond emergency departments, allowing the system to begin looking for signs of sepsis in admitted patients. And it is expected to be housed in UCSD’s newest acquisition, the former Alvarado Hospital Medical Center, now known as the health system’s “East Campus,” on I-8 near the San Diego State University.

Other revisions in the works, Nemati said, include allowing the algorithm to look for more data when there isn’t enough to make a prediction.

“Because COMPOSER has insight into its own uncertainty, it has the ability, when uncertainty is high, to request additional diagnostic testing,” Nemati said. “So a nurse gets a pop-up that says, ‘Hey, this patient is at risk for sepsis, can you run another set of vital signs?’ »

UCSD is also experimenting with advanced wearable patient sensors that could improve the quality of real-time data flowing through electronic patient records, providing more accurate measurements to make predictions. Significant progress, the technologist said, has also been made using broad language models such as ChatGPT, the model that has given every teacher in the world anxiety, to understand doctor’s notes recorded in records patients.

“We’re currently using these notes written at the time of emergency triage to determine what diagnostic hypotheses doctors have in mind,” Nemati said. “This is going to help us reduce false alarms by helping to understand, you know, what other potential explanations are for the anomalies that we’re seeing in patient data.”

UC San Diego Health recently moved forward on several fronts in the field of artificial intelligence, hiring its first director of AI, partnering with Microsoft Inc. to use ChatGPT to help doctors answer common questions from patients and looking for a new hub to help centralize data. to enable deeper integration of AI into frontline patient care.

California Daily Newspapers

Back to top button