Real-Time Physiological Visualization of the Comatose Human Brain

Investigators will create the capacity to graphically display real-time brain physiologic changes in comatose patients to guide urgent clinical interventions to prevent further damage. Continuous multimodal monitoring systems that measure and record changes in brain physiology in comatose patients treated in intensive care units (ICUs) are improving care for patients who have sustained catastrophic brain injuries from trauma, stroke, cardiac arrest and other causes. These brain monitoring devices each incorporate a microprobe inserted through the skull into the brain, which together simultaneously reports on dynamic and interactive changes in cerebral blood flow, brain tissue oxygenation, other brain chemistry, and EEG waves deep in the brain, and on effects of various clinical interventions on these interactions. With prior Dana support, Columbia University's New York Presbyterian Hospital investigators developed one the of the first functioning ICU systems. Instead of intervening once a problem arises, clinicians are beginning to be able to detect physiological derangements that lead to secondary brain injury, intervene to prevent or avert these, and associate physiological and neurochemical measurements with outcomes. To fully transform care, though, the next generation system needs to automatically analyze patient data in the background and present real-time physiologic visualizations of the comatose brain that simplify the clinician's rapid interpretion of massive amounts of data required to make optimal treatment decisions. To develop this unified system, the Columbia investigators will work in a consortium of ten medical centers led by the non-profit Draper Laboratory, using an IBM data streaming analysis platform that handles millions of physiologic events per second. A primary analysis to be tested is adapted from one used to discern the sea surface temperature fluctuations (in locations in time and space) that lead to El Niño. The analytic approach, called empirical orthogonal function analysis (EOF), ranks the proportional mixture of features by the amount of variance that each feature explains thereby serving as a data reduction method and highlighting the most important physiologic features from over 200 variables. The Columbia investigators will create a visualization decision support dashboard, the bedside ICU monitor of the future. The system will create early warning tools and decision support systems to improve outcomes by: 1) identifying physiologic end points as targets for goal-directed brain therapies; 2) identifying onset of secondary complications and the ability to monitor response to interventions; and 3) predicting early harmful physiologic process that can affect overall clinical management. The tool will be tested against other sources of data from 20 comatose patients to assess its reliability and validity. Additionally, analyses of aggregate data are anticipated to detect and identify physiological patterns, leading to experimental clinical strategies that could be tested in clinical trials. Several federal funders have expressed interest in potential large-scale funding if pilot studies indicate that it can lead to improved outcomes and reduced costs. Ultimately, the investigators anticipate that this next generation system will be adopted by health care device manufacturers for widespread use in ICUs.