A predictive algorithm developed by Helios Diagnostics has cleared regulatory review after a 14-hospital trial showed it could flag sepsis onset in ICU patients up to six hours before clinical symptoms appeared, giving care teams a critical window to intervene.
The model ingests continuous vitals, lab draws, and nursing notes, scoring deterioration risk every fifteen minutes. In the trial cohort of just over 9,000 patients, sites using the alert system saw an 18% reduction in sepsis-related mortality and a measurable drop in average ICU length of stay.
Unlike earlier scoring systems built on static thresholds, the model continuously recalibrates against each hospital’s own patient population, which researchers say explains much of the accuracy gain over legacy early-warning tools.
Rollout will begin with academic medical centers that participated in the trial, with a broader release to community hospitals planned for next year pending integration work with existing EHR vendors.
“The hard part was never detecting sepsis once it’s obvious,” said the trial’s lead clinical investigator. “It’s catching the six hours before it’s obvious — that’s where we actually save lives.”