Sepsis kills at least 350,000 Americans annually, yet doctors often can't identify it until it's too late. Now two major healthcare systems are betting on artificial intelligence to change that timeline. Sinai Hospital in Maryland has begun using Sepsis ImmunoScore, the nation's first FDA-authorized AI diagnostic tool for sepsis, while Vanderbilt Health is leading a nationwide platform designed to accelerate testing of next-generation critical care technologies.

Why Is Sepsis So Hard to Diagnose?

Sepsis is a medical emergency where the body's immune response to infection damages its own tissues and organs. Every year, at least 1.7 million adults develop sepsis in the United States, and one in three hospital deaths involve sepsis at some point during the stay. The problem is that sepsis mimics many other serious illnesses, especially in its early stages. There is no single test that immediately confirms sepsis, so physicians must rely on clinical evaluation and laboratory findings while blood cultures are processed, a process that can take several days.

That delay matters enormously. "Sepsis is notoriously difficult to diagnose early because its signs can mimic many other serious illnesses," explained Dr. Gregory Corcoran, chief of emergency medicine at LifeBridge Health. "Anything that helps physicians identify patients at risk sooner can make a meaningful difference in how quickly treatment begins".

Dr. Gregory Corcoran, chief of emergency medicine at LifeBridge Health

How Does the New AI Tool Work?

Sepsis ImmunoScore analyzes up to 22 clinical and laboratory measurements to identify patterns that may indicate a patient has sepsis or is at risk of developing it within 24 hours. The tool is integrated directly into Sinai Hospital's electronic medical record system, meaning clinicians don't need to take extra steps to access results. Instead of creating additional work, the AI generates a risk score that helps clinicians assess whether a patient needs closer monitoring or more aggressive treatment.

What makes this tool particularly valuable is its potential to reduce unnecessary antibiotic treatment while catching true sepsis cases earlier. "By providing additional diagnostic confidence early, it helps physicians make more informed decisions as patients transition from the ED to inpatient care. That can reduce unnecessary antibiotic treatment for patients who don't have sepsis while helping us identify those who do before their condition worsens," noted Dr. Corcoran.

Dr. Corcoran

From Research Lab to Hospital Bedside

Sinai Hospital played a crucial role in developing this technology years before it received FDA authorization. Sinai was one of 10 research centers nationwide that contributed patient samples to a database of more than 100,000 samples used to develop and validate Sepsis ImmunoScore. This collaboration between researchers and clinicians demonstrates how real-world data from actual patients can drive AI development in healthcare.

LifeBridge Health, which operates Sinai Hospital, now plans to evaluate the technology's impact on patient outcomes, including treatment decisions, length of hospital stay, mortality rates, and readmissions. The health system may also expand its use to other hospitals within the LifeBridge network.

A Broader Push to Speed Up Critical Care Innovation

While Sinai focuses on deploying an existing AI tool, Vanderbilt Health is taking a different approach: building an entire platform to test and validate new critical care technologies faster. Vanderbilt has been awarded a federal contract providing up to $11.2 million over five years to establish the Goldilocks Program, which stands for Groundbreaking Organization Leveraging Data Insights: Learning for Optimized Critical care Knowledge and Systems.

The program is part of a larger federal initiative called CIRCLE, the Critical Illness Immunological Reprogramming and Control Point Learning Engine, funded by the Advanced Research Projects Agency for Health (ARPA-H), an agency within the U.S. Department of Health and Human Services. Each year, more than 4.6 million people in the United States are treated in hospital intensive care units for critical illnesses and injuries that can trigger life-threatening immune and inflammatory responses.

"CIRCLE represents an unprecedented opportunity to fundamentally change how we understand, monitor and treat critical illness," said Wesley Self, MD, MPH, Senior Vice President for Clinical Research at Vanderbilt Health and project lead for the award.

Wesley Self, MD, MPH, Senior Vice President for Clinical Research at Vanderbilt Health

How Can Hospitals Speed Up Critical Care Innovation?

The Goldilocks Program aims to dramatically compress the timeline for moving new health technologies from development into real-world testing. Currently, it takes 12 to 18 months to move new health technologies from the readiness stage to first-in-patient testing. Goldilocks is designed to reduce that timeline to fewer than 90 days by providing continuous feedback, testing, validation and clinical guidance throughout development.

• Simulated and Real-World Testing: Investigators will assess technologies in both simulated ICU environments and actual hospital settings to ensure they work in practice, not just in theory.
• Adaptive Clinical Trials: The program will conduct flexible clinical trials that can adjust based on emerging data, rather than following rigid protocols that may miss important signals.
• Regulatory Guidance: The platform will generate evidence suitable for regulatory review and transition to commercial adoption, helping promising innovations move from development to patient care more quickly.

The Goldilocks Program brings together nationally recognized expertise in critical care medicine, clinical trials, biostatistics, implementation science, artificial intelligence, biomedical informatics, commercialization and regulatory science. It will engage a network of up to 25 participating institutions and more than 100 intensive care units across the United States to support product testing, validation studies and adaptive clinical trials.

What Does This Mean for Patients?

These two initiatives represent a shift in how healthcare systems approach AI adoption. Rather than waiting for perfect algorithms developed in isolation, hospitals are now testing AI tools in real clinical environments where they must prove their value against the complexity of actual patient care. Sepsis ImmunoScore at Sinai Hospital demonstrates that FDA-authorized AI can integrate seamlessly into existing workflows, while Vanderbilt's platform shows how to systematically accelerate the development of next-generation critical care innovations.

"Critical illness remains one of the most complex and deadly challenges in medicine," emphasized Dr. Self. "This award positions Vanderbilt and our national partners to help create a new era of precision critical care that is smarter, faster, and more effective for patients and families".

Dr. Self

The convergence of these efforts suggests that healthcare AI is moving beyond the pilot phase. Hospitals are no longer asking whether AI can help in critical care; they're asking how to deploy it effectively and how to accelerate the development of the next generation of tools. For patients with sepsis and other life-threatening conditions, that shift could mean the difference between early intervention and irreversible organ damage.