From descriptive to diagnostic analytics for assessing data quality: An application to temporal data elements in electronic health records

Abstract

Electronic health records (EHRs) offer an alternative rich source of data for research. However, additional steps are required before utilizing them for research purposes. One of the important steps is the assessment of data quality. In this paper, we illustrate an objective and quantitative framework for assessing data quality of temporal data elements and identifying sources contributing to the variation in data quality. The identified sources of variation in our application can be attributed to the implementation of the EHR system. A close collaboration between analytics and domain experts is pertinent to ensure a holistic assessment of data quality. The approach proposed in this paper demonstrates the potential to advance the assessment of data quality from descriptive to diagnostic analytics in a clinical setting by utilizing data mining techniques on features generated from EHRs in a data rich era.