'Reference' as standards

Abstract

In the era of big data, the scientific and social demand for quality data is aggressive and urgent. This dissertation sheds light on the expanded role of reference standards as a mode of scientific and social coordination for identifying data uncertainty and managing data quality through the expansion of the philosophy and language of metrology. In this regard, this study explores the mechanism of the national standard reference data (SRD) program of Korea, which is a unique legal and institutional framework for data certification. As a useful data evaluation approach, this study suggests the concept of ‘data traceability’ with the ‘matrix of data quality evaluation’ according to the elements of a data production system and related evaluation criteria. In particular, this research examines two national SRD projects for data quality assurance of brain magnetic resonance imaging data and national health screening blood glucose measurement data. The study describes how various social and scientific elements are reconfigured and assembled to manage uncertainties inherent in health and medical big data and generate a set of national standard reference data for stroke diagnosis and diabetes management. The two different forms of standards networks involved in the national SRD projects explored in this dissertation reveal the realities and difficulties regarding medical data quality assurance and suggest alternative forms of standards governance by redefining the identities of medical specialists and by creating a special clinical laboratory quality auditing scheme. This allows for various types and levels of knowledge, experience, and techniques to compete and be assembled horizontally for the purpose of identifying the known and suspected uncertainties and defining a socially or technically recognized ‘reference’ as standards to ensure a stated level of confidence regarding data quality within a given context. In conclusion, this study emphasizes the important role of standards governance, which extends beyond a means of regulation and serves as an efficient means of building ‘networks’ or ‘relationships’ between social and technical elements necessary to create a common language and method for the purpose of managing the complexity and uncertainty of big data.