AI models predict sepsis in children

2 days ago

Copy Link

Sepsis is a leading cause of death in children worldwide.
AI models developed by Northwestern University and Lurie Children’s Hospital can predict sepsis risk within 48 hours using EHR data from the first four hours in the emergency department.
The study is the first to use AI models based on the new Phoenix Sepsis Criteria.
The predictive models aim to balance accurate identification of at-risk children without overdiagnosing those not at risk.
The study used retrospective data from 2016-2020 and validated the models with 2021-2022 data.
Predictive features included triage scores, vital signs, and pre-existing conditions like cancer.
Future research will combine AI models with clinician judgment to improve predictions.

Hasty Briefsbeta