Materials and data generated and analysed to verify level of overlap in indication information from DrugCentral, LabeledIn and InContext. Currently we have manually curated indications for 150 anti-cancer and cardiovascular drugs using the Hypothes.is webpage annotation tool to highlight indications in digital drug labels on the DailyMed website - we call our curation InContext. We then compare InContext indications to the ones specified by DrugCentral and LabeledIn.

1. Analysis-figures/
   Figures in .png format demonstrating the number of overlapping indications in our analysed datasets

2. LabeledIn-comparison/labeledin_crowd.dsv
   Official LabeledIn dataset (portion which was crowdsourced)

3. LabeledIn-comparison/labeledin_normal.dsv
   Official LabeledIn dataset (portion which was curated by human experts)

4. LabeledIn-comparison/ManualCuration+DrugCentral.csv
   Manually curated indications and those from DrugCentral for the 150 drugs in this study

5. LabeledIn-comparison/LabeledInParser.ipynb
   Jupyter notebook (Python) generating LabeledIn_comparison_results_table.csv by automatically analysing labeledin_crowd.dsv, labeledin_normal.dsv and ManualCuration+DrugCentral.csv

6. LabeledIn-comparison/LabeledIn_comparison_results_table.csv
   The results comparing an existing drug-indication database (LabeledIn) to DrugCentral and our manually curations (file is generated by LabeledInParser.ipynb)

7. MainData.xlsx
   Main data file with indications from DrugCentral, our manual curations, and analyses by medical experts (and computer algorithms by Ted Pedersen) to determine semantically similar (equivalent) indications that have different names

8. InContext Annotation Protocol.docx
   Document fully detailing step-by-step protocol for annotating drug product labels on DailyMed with therapeutic usage information.