Data driven multivariate measure of data quality in patient records: applications of a novel fast filter approach

M. Wildi,Jürg Dietrich,D. Zahnd
DOI: https://doi.org/10.4414/SMI.23.135
2007-04-13
Abstract:Summary Attempts to utility- or performance-measurements in the hospitalization sector are heavily dependent on the quality of the available data. Therefore, a robust indicator for the overall data quality is needed. We here propose a new and fast multivariate quality-check procedure that improves upon rule-based (validator software) and statistical (general linear) models. Interpretability of suspicious entries in data sets is enhanced by an asymmetric combination of particular data-dimensions. In certain circumstances the proposed multivariate data-screening may be combined favorably with simple (for example bivariate) deterministic rules.
What problem does this paper attempt to address?