Contribute Media
A thank you to everyone who makes this possible: Read More

A Machine Learning Approach to Protect Patients from Blood Tube Mix-Ups

Description

A wrong blood-in-the-tube (WBIT) error occurs when blood collected from one patient is labeled as though it was collected from a different patient. While rare, these errors can cause serious, potentially life-threatening patient safety events. This talk is about how a team of pathology informaticists and data scientists developed and deployed a multi-analyte WBIT detection model at the Children's Hospital of Philadelphia. We describe how machine learning models can potentially identify previously undetectable WBIT errors and improve upon the current detection methodology. Furthermore, we demonstrate how using R markdown, tidymodels, vetiver, and Posit Connect allowed for rapid model iteration, reproducibility, deployment, and monitoring.

Details

Improve this page