Data Integration for the Study of Outstanding Productivity in Biomedical Research

Abstract

Our goal is to analyze improvement of scientific performance in a multidimensional outcome space, with a focus on American biomedical research. With the growing diversity of research databases, limiting assessment of scientific productivity to bibliometric measures such as number of publications, impact factor of journals and number of citations, is increasingly challenged. Using a wider range of outcomes, from publications through practice improvements to entrepreneurial outcomes, overcomes many current limitations in the study of research growth. However, combining such heterogeneous datasets raise three challenges: 1. gathering in one common place a variety of data shared as csv, xml or xls files, 2. merging and linking this data, that sometimes overlap, 3. inferring the impact of inclusive practises, that are often missing from the datasets. We would like to present our solution for the first of those challenges, and discuss our leads for the second and third challenges.