Repeat: a framework to assess empirical reproducibility in biomedical research

Area of Science:

• Biomedical data science
• Health informatics
• Research methodology

Background:

• Reproducibility is critical for scientific rigor.
• Concerns about reliability in biomedical research are increasing.
• Efforts are underway to improve research transparency and accessibility.

Purpose of the Study:

• Develop an assessment tool for research transparency in biomedical secondary data research.
• Operationalize key concepts of transparency using electronic health record data.
• Create the Reproducibility and Transparency Assessment (RepeAT) tool.

Main Methods:

• Developed RepeAT through a multi-phase process.
• Coded and extracted best practices for reproducibility from publications.
• Field-tested the instrument and refined variables.

Main Results:

• RepeAT comprises 119 variables across five categories.
• Preliminary testing on 40 manuscripts showed strong inter-rater reliability.
• Identified areas for further research and refinement of RepeAT.

Conclusions:

• RepeAT can enhance understanding of current research transparency practices.
• Adoption of RepeAT may improve reporting and availability of research outputs.
• Facilitates comparisons of transparency across domains and institutions.