A Statistical Approach for Visualizing the Quality of Multi-Hospital Data

Abstract

The age of Big Data and the associated proliferation of large data sets have necessitated the development of methods that allow for an easy interpretation of data analysis results. Such methods are usually the product of a symbiotic relationship between the ?elds of data visualization, infographics, and statistics. In this work we explore the interplay between data visualization and the mathematical framework used to analyze inter-hospital di?erences in database queries. Such di?erences can re?ect disparities in the quality of care or more fundamental disparities in data quality. As the volume of queries is large and increasing, it is important to develop an incisive way of visualizing these di?erences. Speci?cally, we demonstrate the importance of choosing a mathematical framework that calculates the statistics necessary to visualize the results in a maximally concise and intuitive way. We derive symbolic statistical representations of inter-hospital query di?erences using a Bayesian probabilistic formalism to indicate statistically signi?cant discrepancies. These statistical representations serve the need for visual representation of differences and their meaning apart from statistical expertise. The calculations were performed with a publically-available package, DQM, available at http://sourceforge.net/projects/databasequalitymanagement.