Using a Machine Learning Tool for Blood Culture Decisions in Adults Visiting the Emergency Department A Randomized Trial

What this Trial Is About

This research aims to evaluate whether using a blood culture prediction tool based on machine learning is as effective as current practices for adult patients in the emergency department who need a blood culture analysis. The study focuses on patients who have a clinical indication for blood cultures as determined by their treating physician. The primary goal is to compare 30-day mortality rates, with additional outcomes including hospital admission rates, in-hospital mortality, and length of hospital stay. Participants will be randomly assigned to one of two groups. In the intervention group, doctors will use the blood culture prediction tool to decide whether to perform the blood culture test if the predicted chance of a positive result is less than 5%, the test will be cancelled. In the control group, all patients will undergo blood culture analysis as usual. This randomized controlled trial aims to see if the prediction tool can reduce unnecessary blood cultures and associated issues without increasing risks. During the study, patients will be monitored for 30 days to assess mortality, with hospital admission, in-hospital mortality, and length of stay tracked up to 90 days. Data such as laboratory results and vital signs are collected to support predictions. Safety and outcomes will be carefully reviewed to determine if the tool offers a non-inferior approach compared to standard blood culture testing. Participation duration and follow-up will be based on these timelines to capture relevant clinical outcomes.

Appropriate Use of Blood Cultures in the Emergency Department Through Machine Learning