Researchers increasingly are optimistic about the potential scalability of AI screening tools for preventing septic shock among long-term care residents.
Early-sepsis detection software already is out on the market, and AI could help bolster those tools, because signs of sepsis are easily learned by the tech, Johns Hopkins Professor Suchi Saria, PhD, said in a recent interview published in JAMA.
“If you can leverage that rich clinical context [of sepsis] — almost 200, 300, 400 variables — you can start to model based on the context of this patient and you can start to learn far more precise markers than you can do otherwise,” Saria said in the JAMA interview. “Machine learning and AI may be very well positioned to help us solve this problem of early identification, broadly and for sepsis in particular.”
Sepsis, an extreme and often fatal bodily reaction to illness, occurs more often in older adults, and the elderly also are more likely to die from septic shock, studies show.
Overall, one-third of those who develop sepsis end up dying from the reaction, according to the Sepsis Alliance.
The vulnerability of older adults to infections, as demonstrated during the pandemic, makes them prone to illnesses such as gastroenteritis, influenza and the common cold, all of which can escalate into sepsis, the Sepsis Alliance warns.
AI could help caregivers by determining which residents are at a higher risk for both illness and sepsis, Saria said.
The idea that AI could help serve as an early warning system for sepsis has been proven over a handful of clinical studies, including some for which Saria was a co-researcher.
Despite this optimism, Saria emphasized the need for federal regulation of AI in healthcare. Such oversight would contribute to increased transparency and reducing errors. The White House recently issued an executive order on AI that would create new regulatory agencies for this purpose.
