Embedding data provenance into the Learning Health System to facilitate reproducible research

In: Curcin , V 2017 , ' Embedding data provenance into the Learning Health System to facilitate reproducible research ' , Learning Health Systems , vol. 1 , no. 2 . https://doi.org/10.1002/lrh2.10019

Abstract

Our world is increasingly driven by data. Medical, economic and political decisions are made based on automated analysis of ever-growing volumes of data, be they patient treatment decisions generated from rule models or stock trading decisions made by micro-trading tools. Scientific discovery is now all but impossible without data-intensive infrastructures, which have transformed both how science is done and what science is done. The Learning Health System (LHS) community has taken up the challenge of bringing the complex relationship between clinical research and practice into this brave new world. At the heart of the LHS vision is the notion of routine capture, transformation and dissemination of data and knowledge, with various use cases, such as clinical studies, quality improvement initiatives and decision support, constructed on top of specific routes that the data is taking through the system. In order to stop this increased data volume and analytical complexity from obfuscating the research process, it is essential to establish trust in the system through implementing reproducibility and auditability throughout the workflow. Data provenance technologies can automatically capture the trace of the research task and resulting data, thereby facilitating reproducible research. While some computational domains, such as bioinformatics, have embraced the technology through provenance-enabled execution middlewares, disciplines based on distributed, heterogeneous software, such as medical research, are only starting on the road to adoption, motivated by the institutional pressures to improve transparency and reproducibility. Guided by the experiences of the TRANSFoRm project, we present the opportunities that data provenance offers to the Learning Health System community. We illustrate how provenance can facilitate documenting 21 CFR part 11 compliance for FDA submissions and provide auditability for decisions made by the decision support tools and discuss the transformational effect of routine provenance capture on data privacy, study reporting and publishing medical research.