Objectives: Biomedical data competitions are interdisciplinary live or virtual competitive events that require participants to analyze big biomedical data and develop prototype solutions to real-world problems that accelerate innovation of medical applications, improve healthcare technology design, and help streamline healthcare business models. These competitions, when guided by librarian instruction, support participant's understanding of research data management and health data literacy-related competencies. Participation in biomedical data competitions helps participants identify, consider, and develop solutions to challenges associated with how big biomedical data influences patient well-being. Data competitions are particularly useful for highlighting issues associated with the underuse of uniform data standards and issues associated with navigating siloed data. Long-term data competitions allow participants to submit their work in iterative phases and to receive feedback throughout the hacking process. This approach welcomes beginners and attracts participants who might not participate in traditional data competitions. However, these competitions have high attrition rates. As librarians continue hosting long-term datathons and hackathons, we seek to understand participant motivation and attrition factors.
Methods: We hosted 3 three-week-long data competitions in the academic year 2019-2020. One event was team-based, with each teach assigned a team leader; the other two required participants to work individually. Registration for these events included a pre-assessment and a demographic analysis. At the end of each week, participants checked in and received project feedback from judges. In-person interviews were conducted at the end of each competition.
Results: Reported factors that influenced attrition included lack of time, lack of ability, and lack of understanding. These factors were most commonly reported amongst participants competing in individual competitions. Retention rates, reported learning gains, and participant motivation greatly improved for individuals competing in the team-based challenge, where the team leader was often cited as a mentor and important to the team's success
Conclusions: Findings suggest that retention in long-term data challenges is improved by a team-based model, as opposed to an individual model. We plan to further test these findings in AY 2020-2021 by hosting and comparing individual and team-based challenges. Findings will inform the creation of an OER toolkit to support librarians in planning inclusive data competitions. We will also explore another concept revealed by this study, the influence of the team leader as a mentor. This presentation is supported by the Network of the National Library of Medicine Greater Midwest Region under cooperative agreement number 1UG4LM012346. The content is solely the responsibility of the author.