Making human derived data FAIR: feedback from NCI office of data sharing workshop

The US federal government is committed to maximizing its return on biomedical research investment. This tenet is exemplified by policies that promote broad, responsible sharing of research products generated with public funds, including the recent National Institutes of Health (NIH) Final Data Management and Sharing (DMS) Policy and the NIH Updated Public Access Policy. Scientific data management and sharing must occur in a FAIR (findable, accessible, interoperable and reusable) manner for it to be broadly usable and thereby most impactful [1,2]. The NCI Office of Data Sharing (ODS) conducted a series of workshops to identify clinical features and profiles derived from patients and study participants that inform these respective basic, translational, clinical, and populational science research analyses. The workshop outputs lay the groundwork for developing best practice recommendations on high-value, impactful clinical data features to collect and share with the wider research community. The workshop also highlighted additional data types and methodologies represented across the NCI-funded research portfolio that need structured outputs to be better defined to similarly allow broad sharing in a FAIR manner. In this article we summarize the workshop discussions on data use challenges, present potential solutions, and outline attendee consensus on the minimum patient-derived clinical information needed to complete a wide spectrum of cancer research. We further propose preliminary guidance for policymakers and researchers to implement regarding collection and management of human derived clinical data in consistent and impactful ways that can improve the sharing process and outcomes for data end users.