Background: Gaining overview on a biomedical field of research becomes challenging due to the increasing amount of publications. Computer-supported discourse analysis based on bibliometric data could help scientists to identify focal points and trends.

Objectives: The project aimed at implementing an automatic processing pipeline starting from a PubMed query and leading to a suitable visualization of the thematic evolution of a domain fostering meaningful interpretation.

Methods: The four-step processing pipeline includes bibliographic data acquisition, preprocessing, network-based analysis of co-occurring keywords (degree and betweenness centrality), and its visualization by heatmap diagrams.

Results: Applying the implemented workflow we analyzed the field of clinical decision support systems based on 5.094 PubMed results yielding a total of 174.663 inter-keyword links. The resulting heatmap shows e.g. increased relevance of electronic health record and (patients') age.

Conclusion: Network-based discourse analysis can be implemented using an efficient processing pipeline and may add valuable insight on the thematic evolution of scientific domains.