Visualization of multivariate time-series data in a neonatal ICU

Existing visualizations in the neonatal intensive care unit (NICU) frequently obscure important trends in clinical data presented to the clinician in tabular displays or stacked univariate plots of variables as a function of time. Scales and alarm limits in clinical displays are based on data that is typical for adults (i.e., adult “norm data”), resulting in confusing or misleading displays in the NICU. In premature infants, norm data differs significantly both from adult values and among infants of differing gestational ages. Interfaces designed to display adult values hinder the perception of clinical changes. We developed a visualization that provides an integrated, multivariate interface for representing laboratory and physiological data in the NICU. We present its design and evaluation and discuss potential future applications of this visualization that is interactive, animated, and personalized to an individual patient so that clinicians can quickly and efficiently recognize significant changes in the patient's condition.