Improving intensive care unit effectiveness and efficiency through improved data processing and analysis techniques.

The purpose of this project is to make use of the extensive data archiving, searching and analysis techniques for large datasets in the field of astrophysics and apply them to the hospital Intensive Care Unit (ICU) setting. Over the past 10 years in particular there has been a big push to ensure that complex, multi-wavelength, multi-source astrophysics datasets are in a compatible formats to enable astronomers to explore and resources from around the world, find data, store and share files, query databases, plot and manipulate tables, cross-match catalogues, and build and run scripts to automate sequences of tasks. These techniques are used for spacecraft datasets based at MSSL. This alongside with a similar push from the particle physics world has lead to a core expertise within STFC funded research in dealing with large and complex datasets that is the foundation of much of our (astronomy) scientific research today. One example is the Hinode spacecraft data. In this case a user can probe the archive searching for changes in different parameters such as intensity, velocity, turbulence, and view the data in multiple formats such as movie format and lightcurves. The system was designed to be flexible, using standard formats that can be probed by anyone in the world. We have techniques to look for triggers to large explosions on the Sun, which involve gradual changes that lead to a catastrophic outcome. Similar techniques could be used to acquire patient data and track trends in a patient's condition in an intensive care unit (ICU) setting (e.g. heart rate, blood pressure, blood oxygenation, electrolyte disturbances etc.). Currently most of the datasets are taken in a discontinuous way and in many cases are still recorded by hand at 30 minute intervals. The data is not currently stored and cannot be used to determine precursors. The longest period data will be kept in monitors will be on circular buffers for 24 hours with discontinuous data being collected on 30 min intervals to allow longer term study. The ultimate goal of this study is to provide a software tool that will allow a response BEFORE a catastrophic change happens physiologically rather than after, as is currently the case.