The recent increase in accessible medical and clinical laboratory "Big Data" has led to a corresponding increase in the use of machine-learning tools to develop integrative diagnostic models incorporating both existing and new test data. The rise of direct-to-consumer (DTC) testing paradigms raises the possibility of predictive models that use these new sources. This article discusses several distinct challenges raised by the DTC approach, including issues of centralized data collection, ascertainment bias, linkage to medical outcomes, and standardization/harmonization of results. Several solutions to maximize the promise of machine-learning data analytics for DTC data are suggested.
Keywords: Big Data; DTC; Direct-to-consumer testing; Harmonization; Laboratory medicine; Machine learning.
Copyright © 2019 Elsevier Inc. All rights reserved.