Discovery of signatures of fatal neonatal illness in vital signs using highly comparative time-series analysis

Abstract

ObjectiveTo seek new signatures of illness in heart rate and oxygen saturation vital signs from Neonatal Intensive Care Unit (NICU), we implemented highly comparative time-series analysis to discover features of all-cause mortality in the next 7 days. DesignWe collected 0.5Hz heart rate and oxygen saturation vital signs of infants in the University of Virginia NICU from 2009 to 2019. We applied 4988 algorithmic operations from 11 mathematical families to random daily ten-minute segments. We clustered the results and selected a representative from each, and examined multivariable logistic regression models. SettingNeonatal ICU Patients5957 NICU infants; 205 died. Measurements and main results3555 operations were usable; 20 cluster medoids held more than 81% of the information. A multivariable model had AUC 0.83. Five algorithms outperformed others: moving threshold, successive increases, surprise, and random walk. We computed provenance of the computations and constructed a software library with links to the data. ConclusionsHighly comparative time-series analysis revealed new vital sign measures to identify NICU patients at the highest risk of death in the next week.