Skip to main content

Three recent lessons from applying text mining in Medical Affairs  

Text mining medical affairs

In our work with a wide range of medical affairs teams, we have delivered data, insights and reports using text mining approaches to speed up project delivery and unlock new findings. Here are three key lessons we’ve learnt recently:  

  1. Start with the question – During a recent project, we began by capturing key questions and assessing them to understand which would be best suited to automation with Natural Language Processing (NLP), which were best answered by the insights team or a combination of both. This gave us faster results to all questions and a natural priority for which outputs were selected for automation, allowing these to be refreshed on a weekly basis..  

  2. Tune the terminology – When extracting topics and data points from within documents and text rich sources, we start with pre-built ontologies and libraries of terms. These are continuously refreshed and help us identify the information we need and extract it quickly. For more precision, we tune for therapy area specific terms and concepts, helping us reduce noise further and, thanks to the breadth of linguistic and machine learning approaches we use, we can get precise results without large data sets to train on.   

  3. Context is key – Text mining for specific topics and terms within documents can be powerful but tagging these outputs with contextual fields makes the resulting analysis far richer. For example, we recently extracted positive outcomes of a particular therapy from within published literature and pre-prints that were based on real-world evidence sources. Along with these extracted topics, we also extracted date, country, authorship, publication type, patient profile terms and sentiment. With all this information to accompany the core data points, we were able to create a series of analyses from time series to topic heat maps to help the team understand where these key concepts appear, in what context and how they vary over time.   

There is huge potential in the application of NLP for Medical Affairs, and we are excited to be helping teams inform their strategy, plan new research and engage physicians and patients more effectively thanks to a more data driven approach. For more information contact us at  

Visit the Insights Hub page

Ready to get started?

Request a Demo

Questions? Ask our experts