The term open science describes a broad effort aimed at making scientific processes, data, analyses, and publications as accessible as possible. By eliminating barriers to sharing data and improving access to research findings, the open science movement seeks to accelerate learning and discovery worldwide while ensuring that everyone can benefit equally from scientific knowledge.
But open science is about more than just improving access to information. In an important sense, practicing open science can also enable better science. With many different eyes scrutinizing data, new insights may be brought to bear on a problem and diverse perspectives and experiences may help spotlight flaws in methods or interpretations. Applying the principles of open science can ultimately improve the practice and culture of research by making it more accountable, reproducible, and transparent.
Recently, the global community of researchers and scholars has seen the cause of open science and data sharing grow from a relatively niche concern to a major force in the national and international scientific enterprises, spanning academia, government, and industry. Duke University has been an active participant in these efforts, and in recent years multiple open science projects have taken root and flourished across campus.
Surveying the Landscape of Open Science and Data Sharing at Duke
A little over a year ago, the School of Medicine convened faculty and staff from across the University to craft a campus-wide framework for supporting open science and data sharing—including studies, such as clinical trials involving human participants, that require special attention to issues related to data privacy and security. The effort would also need to harmonize with larger national and international efforts while at the same time establishing a clear academic and career incentives for practicing open science at Duke.
Exemplary Elements for Data Sharing
- To be useful for the research community, it’s important to consider methods that facilitate future use, including by investigators who may be unfamiliar with the data and the context surrounding their creation.
- Data sharing plans for studies that involve human subjects must protect privacy and ensure that data are (and remain) de-identified.
- Regardless of the data sharing platform used, discovering data via a central location or library (such as the Discover Data@Duke catalog) will be critically important.
- Data sharing should not limit personal research interests or the development of new ideas by an investigator.
Starting in 2018, this group embarked on an evaluation process comprising a landscape analysis and inventory of potential data sharing efforts and platforms both within and beyond Duke, including the Duke Data Service, Vivli, and Duke SOAR, among others. The group then developed a preliminary set of guideline goals and strategies for open science at Duke and solicited feedback and input through a series of meetings and roundtables, culminating in a February 2019 meeting hosted by Duke at the National Academy of Medicine in Washington, DC.
As efforts to develop a roadmap for open science continue, we recognize that such an effort will only succeed if it embodies a shared mission with shared results. In that spirit, we are seeking input from the Duke community about how we can best achieve our goals. The items listed below reflect some of our current thinking about how we can effectively foster a mature culture of open science at the School of Medicine. As the inaugural Duke School of Medicine Research Week progresses, we’d like to open the conversation by understanding key elements the community deems important and ensuring that our efforts are informed by the best ideas from all stakeholders.
Overall Goals for Open Science
To advance science by:
- Answering multiple new questions
- Combining data to increase power
- Increasing speed of discovery
- Avoiding duplication of efforts
- Fostering new connections and collaborations
To improve research integrity by:
- Validating original analyses
- Promoting transparency and trustworthiness
Guiding Principles
- Facilitate appropriate access to research information, with a range of privacy controls tailored to the nature of the study
- Ensure appropriate oversight while minimizing barriers to data access, to prevent against misuse of original data while promoting new discovery
- Maintain utility of data, such that shared data can be used to generate new analyses
- Expect that results of shared data will similarly be shared
- Acknowledge and provide academic credit for those who contribute original data
Next Steps
To achieve these goals and realize the opportunities for open science, we are looking for additional ideas on exemplary practices (see inset) and incentives for data sharing in an academic institution. We will continue to refine our roadmap over the coming year. As we receive input from the community, we will share more details, solicit additional feedback, and continuously test approaches to fostering open science at Duke.
We look forward to hearing from you! Email your best ideas to us at adrian.hernandez@duke.edu or tweet them to @texhern.