Politico, a leading publication known for its in-depth reporting on policy and innovation, profiled Prenosis:
Many algorithms claim to predict sepsis, an extreme immune response to infection that can sometimes be fatal. Prenosis, a Chicago-based health tech company, says it is the first to receive Food and Drug Administration authorization for its artificial intelligence sepsis prediction tool.
It is classified as “de novo,” which means there is no other equivalent device on the market.
What separates Prenosis technology from traditional sepsis alert systems, is that instead of continuously monitoring for sepsis or signs of sepsis, as most algorithms do, Prenosis creates an on-demand risk score for the patient. Doctors must order an assessment as they would any other lab test.
Another unique aspect of this product is that the AI will suggest that a doctor order additional tests if it thinks the data is insufficient to issue a score.
“Maybe you did run this blood test, but it was 18 hours ago. That’s not good enough because the patient’s health could have changed within the last eight hours,” CEO Bobby Reddy said, noting that it’s not always easy for a clinician to quickly see when all their tests were last run.
Also, in a world where AI is largely a black box, Prenosis’ interface is transparent about what data influences its scoring, instead of simply delivering a result.
The company trained its AI on data garnered from 100,000 blood samples from 25,000 patients at 10 different hospitals to help it look for the biological underpinnings of sepsis.
Why it matters: A third of people who die in hospitals have sepsis, according to the Centers for Disease Control and Prevention. Last year, the agency launched a program to help hospitals identify the infection.
The FDA has previously awarded breakthrough status, a designation that allows technology to be expedited through the approval process, to several sepsis algorithms, including Inflammatix’s TriVerity, Mednition’s KATE Sepsis and Bayesian Health’s Targeted Real-Time Early Warning System.
Some sepsis AI is already being used without the FDA’s explicit authorization. Electronic health record system Epic had a sepsis algorithm that came under fire for its questionable accuracy, though it has since been updated.
What’s next: Reddy said a side benefit of Prenosis technology is that it could potentially help keep medical records more complete and up to date, thereby improving other AI systems’ ability to make predictions.
“Because it’s demanding all the right data,” Reddy explained.
Read the story here.