Description
Single-ventricle (SV) heart defects are critical congenital conditions that require invasive procedures during infancy. In the period before stage-2 surgical palliation, SV infants are at elevated risk of critical events and unexpected deterioration. We identified SV ICU admissions between 2015-2018, and collected longitudinal values of physiological variables including vital signs and laboratory tests. We extracted frequent temporal patterns from these data, and trained machine learning classifiers for the early prediction of critical events
Learning Objective: Understand how longitudinal changes in physiological variables can be used to achive early and accurate prediction of patient deterioration in single-ventricle infants
Authors:
Victor Ruiz (Presenter)
Children's Hospital of Philadelphia
Jorge Galvez, Children's Hospital of Philadelphia
Michael Goldsmith, Children's Hospital of Philadelphia
Maryam Naim, Children's Hospital of Philadelphia
Alejandro Lopez-Magallon, Children's National Medical Center
Ricardo Munoz, Children's National Medical Center
Rich Tsui, Children's Hospital of Philadelphia
Presentation Materials: