nlpandmed

Repo contributors: Benjamin Skov Kaas-Hansen [code author], Davide Placido [code review], Cristina Rodriguez Leal [code review]

The repo holds the public version of our full analytic pipeline of the paper Eliciting side effects from clinical notes: language-agnostic pharmacovigilant text mining. Symlinks and other internal files have been removed as they are non-essential for reading the code in the repo and wouldn't port anyway.

Related publications

Kaas-Hansen, BS, Placido, D, Rodríguez, CL, et al. Language-agnostic pharmacovigilant text mining to elicit side effects from clinical notes and hospital medication records. Basic Clin Pharmacol Toxicol. 2022; 1-12. doi:10.1111/bcpt.13773

Scope of study

Develop a machine learning pipeline for safety signal detection in textual data from electronic medical records.

Design

Case-control-like, but not really.

Data sources (all Danish)

• In-hospital medication data
• In-hospital clinical notes from the first 48 hours of admissions

Important notes for running the pipeline:

• The command bash snakeqsub.sh must be run from the bash shell.
• If you happed to automatically start another shell in .basrc, deactivate that to avoid invoking this alternative shell. Otherwise, processes will linger and clog the pipeline.