Validating the GRADY Machine Learning Model for Early Detection of Sepsis and Bacteremia in ICU Patients

What this Trial Is About

Researchers are evaluating the GRADY prediction models, which use machine learning to estimate the risk of gram-negative bacteremia and sepsis in intensive care unit ICU patients. Sepsis is a serious condition with high mortality that needs early diagnosis and treatment. Current methods like blood cultures take time, and existing scoring systems such as SOFA, SIRS, and NEWS2 may not detect sepsis early enough. This study aims to compare the GRADY models with these standard scoring systems to assess their ability to provide earlier and more accurate risk detection and support timely clinical decisions in critical care. The study will prospectively validate the GRADY models by using routinely collected vital signs and laboratory data from ICU patients. These models will be assessed for their diagnostic accuracy and clinical usefulness compared to existing scores such as SOFA, SIRS, and NEWS2. The study also plans to calculate Pitt Bacteremia Scores to explore relationships with GRADYs risk classifications. This observational study will include ICU patients who have had blood cultures taken as part of their routine care. Participants will be adult ICU patients monitored for at least 48 hours, with blood cultures obtained during their stay. Data collection involves reviewing routine clinical and laboratory information and calculating various scores at admission. The main outcome is detecting gram-negative bacteremia within 28 days. Secondary outcomes include measuring SOFA, SIRS, and NEWS2 scores at admission. The study aims to evaluate the models ability to identify high-risk patients early, potentially improving treatment timing and outcomes in ICU care. The study will run until early 2026.

Prospective Validation of GRADY: A Machine Learning Model for Early Sepsis and Bacteremia Detection in ICU Patients