LifeBridge Health's Sinai Hospital of Baltimore is putting an artificial intelligence-powered diagnostic for sepsis into clinical use, bringing a research-backed tool to the patients' bedside.
Chicago-based Prenosis, a biology-based technology company, developed Sepsis ImmunoScore, an AI-powered tool that can spot the early signs of sepsis. The tool analyzes 22 parameters to identify patterns in a patient’s biological immune response, enabling clinicians to detect sepsis, or the progression to sepsis within 24 hours.
Sepsis ImmunoScore received the U.S. Food and Drug Administration's first authorization for software to diagnose sepsis in 2024. Prenosis’ solution was approved through the FDA’s De Novo pathway.
A study published in the New England Journal of Medicine AI found the Sepsis ImmunoScore has high accuracy for sepsis diagnosis and simultaneously predicted critical outcomes, including mortality during hospitalization, length of stay before discharge, ICU admission, need for mechanical ventilation and use of vasopressors.
Sinai Hospital of Baltimore was among 10 research centers nationwide whose patient samples contributed to the 100,000-sample dataset that trained and validated the Prenosis algorithm. LifeBridge Health, which operates five hospitals and specialized care facilities in Maryland, has a long-standing relationship with the company as health system researchers contributed to the foundational study that enabled FDA authorization of the tool.
"Our collaboration with Prenosis began with research; it’s gratifying to see that work evolve into something our clinical teams can use directly. Adding new tools to our fight against sepsis reflects exactly the kind of forward-thinking innovation that defines our approach to patient care," Charles Albrecht III, M.D., Sinai Hospital Chief Medical Officer, said in a statement.
Sepsis is the most expensive and one of the hardest conditions to manage in U.S. healthcare. Each year, at least 1.7 million adults and more than 18,000 children in the U.S. develop sepsis, according to data cited by the U.S. Centers for Disease Control and Prevention. At least 350,000 adults and more than 1,800 children who develop sepsis die during their hospitalization or are discharged to hospice.
Sepsis can be notoriously difficult to diagnose, and delays in recognition and treatment significantly worsen outcomes.
"Sepsis is just a very challenging diagnosis to make. There's a ton of diagnostic uncertainty from the moment of arrival in the emergency department and then even moving forward as the patient goes into the hospital after they're admitted there's still an uncertainty as to whether this patient truly has infection in the blood or not," Gregory Corcoran, M.D., chief of emergency medicine for LifeBridge Health, said in an interview with Fierce Healthcare. "The only way we really know for certain, and even then, it's not 100% certainty, is the results of the blood cultures, which often take days to come back."
This uncertainty can lead to overtreatment, with some patients receiving unnecessary IV antibiotics and contributing to antibiotic resistance, or undertreatment, where subtle cases of sepsis are missed and treatment is delayed, allowing patients to become sicker, Corcoran noted.
Because patients require complex, multidisciplinary ICU care and extended stays, sepsis consumes a massive portion of healthcare budgets.
AI is very uniquely suited to address the complexity of sepsis diagnosis, according to Bobby Reddy, Jr., Ph.D., CEO and co-founder of Prenosis. "Because sepsis is so complicated, and there's so many different biological subtypes of sepsis, it's important to collect enough information and the right information, and then to compile that information using AI to determine the right treatment for the patient."
Prenosis' Sepsis ImmunoScore, built on the company's Immunix precision medicine platform, integrates biomarkers and clinical data to present real-time critical information to help clinicians act sooner, potentially reducing harm to critical organs and improving patient survival, according to the company. It integrates directly into a hospital’s electronic medical record system.
The company took the approach, backed up by research, that biology-driven AI could change how sepsis is diagnosed.
"We took an approach that sepsis is not just one disease," Reddy told Fierce Healthcare. "We need to treat sepsis as a heterogeneous syndrome, not as a single disease, that is made up of a bunch of different types of biology that go into sepsis. In order to better understand this very heterogeneous, complicated syndrome, Prenosis, over the last 10 years, has been building up a storage of biological samples, the blood samples that were taken from patients, along with all of the data that was collected for the patients, so vitals, labs, medications, comorbidities, demographic information, along with the physical blood samples that were taken from the patients, and then we bring those blood samples into our lab in Chicago, where we measure a lot of deeper biological parameters that are not measured by the hospital."
He added, "We're literally talking about hundreds of parameters for each patient, and then we use AI to better break down the different types of biology that make up sepsis. It's truly a precision medicine or personalized approach."
It typically takes less than an hour from the time the diagnostic test is ordered to the result, Reddy said. "The sepsis score gives the clinician an idea of the risk of developing sepsis in the next 24 hours, so it's very predictive," he said.
While antibiotics and other antimicrobial treatments can target the underlying infection, there are currently no therapies that directly treat the immune response driving many sepsis cases.
Prenosis also is developing personalized immunotherapies that tailor treatment based on a patient's specific immune profile, Reddy said.
Sinai Hospital of Baltimore mobilized the AI tool clinically six weeks ago and it has been used, to date, to test or diagnose 75 patients, Corcoran said.
"The reality of this tool is that the information that we are obtaining by ordering this in the emergency department is not typically going to change our care in the ER. However, it sets the patient up potentially for success later on, as there's increased diagnostic accuracy for the hospitalists when they take over care of these patients," Corcoran said. "The key to our success in the emergency department is a lack of impedance of workflow. If it's easy for our team to use, and it's not creating extra steps, and it's not creating extra work for the ER providers, and yet downstream we're getting benefit as diagnostic accuracy, then we know that's good for our patients, and we're more apt to use it."
As Sepsis ImmunoScore is put to use in clinical care, LifeBridge Health will be evaluating several metrics, including ongoing assessment of treatment and care for septic patients, executives said.
RELATED: Prenosis nabs de novo nod for AI-powered sepsis risk tool
Prenosis evaluates how the tool impacts hospital compliance with the SEP-1 bundle (Centers for Medicare and Medicaid Services Severe Sepsis and Septic Shock Management Bundle) while also assessing how the tool improves length of stay, in-hospital mortality and 30-day readmission rates, Reddy said.
Building the infrastructure within Sinai Hospital will lay the groundwork to deploy the AI solution across LifeBridge Health's additional hospitals, Martin Gesheff, executive director of research at LifeBridge Health, said.
The company recently inked a Biomedical Advanced Research and Development Authority (BARDA) contract and the completion of a series A financing round to expand its Immunix platform across conditions including pneumonia, acute heart failure, acute respiratory distress syndrome, and acute kidney injury. Prenosis also recently announced a strategic collaboration with BioAegis Therapeutics to identify patient subpopulations most likely to benefit from targeted inflammatory disease therapies.
"We think there's an opportunity to develop precision drugs for a wide variety of indications across the acute care setting, including sepsis, acute heart failure, pneumonia, acute kidney injury and acute respiratory distress syndrome," Reddy said. "What we've done through building up our biobank and the collection of biological data and clinical data is we've discovered different signatures of biology that span many of these different conditions."
AI will revolutionize healthcare by driving biology-based precision medicine in acute care, Reddy asserted. "I'm excited about the potential of AI to revolutionize this particular problem which is about understanding the complex biology of these patients, so that the right patient can be matched to the right treatment at the right time. The timing piece is important because in the hospital setting the biology of the patient can change rapidly in the course of hours or days," he said. "I think we're going to see a radical transformation over the next five to 10 years of bringing true precision medicine that's enabled by AI."