Screening Newborns for Critical Congenital Heart Disease Using Oxygen and Perfusion Measurements with Machine Learning

What this Trial Is About

Researchers are evaluating a machine learning ML algorithm that combines pulse oximetry features, including oxygen saturation and perfusion measurements, to screen for critical congenital heart disease CCHD in newborns. This study aims to externally validate the algorithm previously tested internally, assessing if adding perfusion measures and repeated measurements improves detection accuracy and reduces false positives. Participants will have non-invasive oxygen saturation SpO2 and perfusion index PIx measured using pulse oximeters. The ML screening algorithm will analyze these measurements continuously, assigning predictions every minute. The study includes repeated measurements up to four times, including after 48 hours of age, which can be done outpatient. This approach supports creating a dynamic model that incorporates new data over time. During the study, newborns suspected of or undergoing screening for CCHD will have their SpO2 and PIx monitored. Researchers will collect data to evaluate the algorithms performance by measuring sensitivity, specificity, and the area under the receiver operating characteristic curve over approximately four years. Safety monitoring and follow-up are integrated into usual care with assessments through the newborn period up to 22 days of age.

Dynamic Critical Congenital Heart Screening With Addition of Perfusion Measurements