Projects

Project Overview

In this work, Insight assisted the Office of Behavioral and Social Sciences Research of the National Institutes of Health (NIH) in a review of the portfolio of behavioral and social science interventions to assess how many studies tested a novel intervention and how many tested an adaptation of an existing intervention. There were two goals of this study. The first goal was to determine the feasibility of using natural language processing to review study abstracts and distinguish studies testing novel interventions from studies testing an adapted intervention. The second goal was to determine whether the study abstracts included sufficient information to assess whether the study was testing a novel or adapted intervention.

NIH provided Insight with research abstracts from funded studies and guidelines for reviewing the abstracts. Overall, Insight coded 744 abstracts and recorded the results. Study findings concluded that most abstracts in the sample did have the information needed to classify the intervention as novel or adapted. However, because the information provided in each of the abstracts varied in content and specificity, the coding and classification would be impossible to automate (through natural language processing, for example).

Products

Series of memoranda describing findings