An early warning system for sepsis, one of the deadliest infections for hospitalized patients, has been approved for use by the FDA, making it one of the first AI-based medical tools to receive clearance.
The tool, developed by researchers at Johns Hopkins University and now commercialized by Bayesian Health, detects sepsis hours earlier than doctors and reduces deaths by almost 20%.
“Pre-suspect screening is what creates lead time, and lead time is what changes sepsis outcomes. Once a clinician already suspects sepsis, the clock continues to tick, often for hours and even days,” says lead researcher Suchi Sarria. He is a professor at Johns Hopkins University and director of the AI and Healthcare Research Institute, who began translating laboratory research into real-world systems after losing his nephew to sepsis in 2017. Sepsis must be monitored with tests or equipment before clinicians suspect it. ”
In sepsis, each hour of delay in sepsis detection significantly reduces a patient’s chances of survival. Sepsis is often overlooked because symptoms such as fever and confusion are common with other medical conditions.
To overcome these possibilities, Thalia and the Johns Hopkins team created a targeted real-time early warning system. A federally funded study that integrates electronic medical records with advanced clinical AI allows doctors to detect cases of sepsis two to nearly 48 hours earlier than traditional methods.
This system has reduced sepsis mortality rates by 18% in dozens of hospitals across the United States. This is a major advance in combating the deadly immune response that claims more than 250,000 lives each year.
It gives doctors extra eyes and ears and could truly help save lives. This is an important milestone for Johns Hopkins University and Dr. Thalia’s team. ”
Albert Ou, Johns Hopkins University Patient Safety Specialist and Study Co-Investigator
In 2023, under the FDA’s Breakthrough Designation (a designation that promotes technologies that have the potential to improve the treatment of life-threatening conditions), the technology was implemented in several health systems, including the Cleveland Clinic, Memorial Care in California, and the University of Rochester School of Medicine, and significantly reduced in-hospital mortality, morbidity, and length of stay for patients with sepsis.
“There are very few clinical AI systems that can reason across messy real-world hospital data and provide guidance that allows clinicians to act reliably,” said Sarria. “The FDA’s approval is the first regulatory authority to change the standard of care for a disease that is associated with approximately one in three hospital deaths. This means that decades of clinical AI research at Johns Hopkins has been put into practice. It means that we are not just models built in the lab, but technology where it matters most: at the bedside.”
FDA clearance also allows hospitals using the system to receive Medicare and Medicaid reimbursement under the New Technology Supplemental Payment Program, which compensates hospitals for the use of new technology.
Suchi’s work has reached a major milestone. They can now play an important role in preventing deaths and complications in hospital. ”
Ed Schlesinger, Dean, Whiting School of Engineering, Johns Hopkins University
