We created a human-in-the-loop method that used text and image analysis to conduct a systematic review of data visualizations from a corpus of research articles. We applied our method to the genomic epidemiology scientific literature and produced a Genomic Epidemiology Visualization Typology (GEViT).
Adjutant supports literature reviews by obtaining, analyzing, summarizing, and visualizing research articles from a PubMed search by performing a fast and unsupervised topic clustering. It is distributed as an R package and can be used through a Shiny-based GUI or as a series of commands within and R script.
We used a multi-phase mixed methods research design to study how public health experts make decisions with routine clinical data and data derived from tuberculosis whole genomes. Using our findings we implemented a clinical report that has been adopted by international public health agencies.