SciDisruptor-PubMed

Research hypothesis and data interpretation
This project is based on the hypothesis that patterns of disruption in scientific publications can be identified in a fully automated and reproducible way through citation networks and the Disruption Index. To test this, a series of Python scripts was developed to extract, organize, and analyze data directly from the PubMed database.

What the code demonstrates
The scripts automate the entire process: (1) querying articles by keyword using the PubMed E-utilities API; (2) extracting complete article data in XML format; (3) cleaning and organizing the extracted data; (4) building the citation network among the selected articles; and (5) calculating the Disruption Index using the formula proposed by Wu, Wang, and Evans, via the PySciSci package.

How to interpret and use the code
Each script is modular and designed to run sequentially, with comments and instructions for execution in a Jupyter Notebook environment. Input data is retrieved directly from PubMed, and the outputs include: PMIDs of all articles, structured citation networks, tables with the count of citation types (A, B, C), and the final Disruption Index score. The code can be used by other researchers to replicate this study or to apply the pipeline to other scientific domains.