Prediction of Persistent and Profound Intraoperative Hypoxemia in Pediatric Patients to Mitigate Life-Threatening Events and Improve Patient Safety
The Opportunity
Refractory intraoperative hypoxemia is defined as low oxygen in the blood during an operation that does not improve by increasing the concentration of inhaled oxygen. Low oxygen levels can have several possible causes, can happen unexpectedly even during routine operations, and are immediately life-threatening, if not properly treated. The goal of this project is to use preoperative and intraoperative data to develop a machine learning algorithm for predicting hypoxemia.
Our Approach
The team worked to fine-tune an existing model and develop and test a “silent” clinical-decision support application using prospective, real-time data. Then, they integrated the Deep Intraoperative Alert (DIA) model framework to detect intraoperative hypoxemia within the Epic Electronic Health Record (EHR) and assess clinical meaningfulness of the alert system.
The Solution
The final report for this project is not complete, so please check back at a later time for more information about the outcomes of this project.
