BrainFlux Project

Intensively sampled longitudinal data are ubiquitous in modern medicine. Hospitalized patients typically have multiple physiological parameters acquired continuously as waveforms with numerous other measures repeatedly sampled over time. Despite the wealth of available information, clinical providers and biomedical researchers typically rely on massively down-sampled data and are limited in their capacity to integrate complex data into clinical decision making.

Our work on the BrainFlux project has the following thrusts:

Data Rescue. Continuous accumulation, storage and aggregation of highly multivariate time-series data into a hierarchy of materialized synopses that facilitate efficient medical decision-making.
Data Analysis. Use of past observations to make precise estimates of patients’ anticipated clinical courses and outcomes. Effective summarization of these complex computational processes to the bedside provider to allow easy interpretation of these predictions.
Data Monitoring. Real-time data monitoring to leverage complex data relationships for the purpose of early detection of signs of instability or side effects. Real-time predictions of outcome based on complex data relationships in highly multivariable time series data.

We focus on a well-defined and challenging patient population: comatose survivors of cardiac arrest. Sudden cardiac arrest is the most common cause of death in high-income nations, with over 600,000 cases in the United States annually. Comatose post-arrest patients undergo advanced, multimodality clinical monitoring resulting in multiple data streams. One of the richest data sources in terms of the treatment goals delineated above is electroencephalography (EEG). EEG measures brain activity and is typically monitoring across 22 channels at 256 Hz. A single day’s EEG recording of one patient results in approximately 4.46x10^8 data points. There is an urgent need for quick and clinically effective summarization and utilization of these complex data, which we address in design and development of advanced BrainFlux data eco-system.

For details, please check our recent papers. Here is a summary presentation at grand opening of House of CAIR.

This page is maintained by Department of Informatics and Networked Systems, School of School of Computing and Information at the University of Pittsburgh