National Trauma Data Bank (NTDB) is the largest repository of statistically robust trauma data in the United States, assembled from trauma centers across the country. NTDB data has been commonly used in risk adjusted studies in the medical communities to describe patterns of injury, interventions and patient outcomes in order to better tailor trauma treatment. The studies have led to significant improvements in the standard of care delivered to trauma patients. A considerable amount of research efforts have been spent on development and maintenance of NTDB to continuously improve the quality and effectiveness of trauma patient records. Prior studies relied mostly on ad hoc and manual extraction processes of data from NTDB repository. Given the rapid growth of the NTDB datasets in an ever changing clinical environment, there is an urgent need to develop standard methodologies and software tools to support data analysis involving NTDB datasets. The goal of this research is to empower clinicians to be able to utilize collected content for such analysis by using standardized data collection and aggregation practices. Specifically, in this paper we generalize existing OLAP techniques to model NTDB data for capturing statistical and aggregated information. We present a system to automate the process of creating ``incident cubes'' for all permutations of attributes in NTDB data model, and a querying framework for extracting information from cubes. We also define a ranking function to discover new and surprising patterns from cubes, based on the information gain from each attribute. A case study is used to illustrate that we can take advantage of the system to support trauma data analysis effectively and efficiently.
展开▼
