To develop a critical care data warehouse to support multi-institutional clinical data management and facilitate knowledge discovery

To test the hypothesis that probabilistic modeling of human physiology and ICU sensor function can improve clinical diagnosis and decision making in critical care

To test the hypothesis that data-driven informatics methods can identify patterns in multivariate clinical data that define patient physiology and predict outcome

Aim 1

Aim 3

Aim 2

Project List: Specific Aims

· Our group is one of a small number of institutions that have created site-specific hardware and software solutions to acquire ICU physiological data. While these early efforts have demonstrated the potential for biomedical informatics in critical care, they have highlighted the need for a large shared knowledge base that can only be achieved through multi-institutional collaboration. They also point out the requirement for automated data filtering methods for quality assurance. We are developing a highly scalable data warehouse using standard hardware and software; when it is complete, we will transfer our ICU data into the warehouse and begin to acquire ICU data from a second hospital. As this data warehouse grows, it will be capable of supporting multi-institutional studies to facilitate biomedical informatics research and ultimately improve patient care.

C-BICC -

Center for Biomedical Informatics in Critical Care