Sorry, we don't support your browser.  Install a modern browser
This post is closed.

Automated warning messages for incorrect PDF uploads#480

I was doing some QC on a nest today and stumbled across a case where the PDF uploaded for a study did not match the actual study that needed screening. Would be really cool if an automated warning message pops up when the title of the uploaded PDF does not match the title for the article to be screened. This occurrence should be rare and study screeners should be careful, but still think this would be a valuable feature that can be quite time saving. An example of this issue is attached

4 years ago

Good call– I think you’re right, not a likely issue, but destructive when it actually happens… could lead to gathering every single tag and data element incorrectly.

The feature request is super simple here; title matching between our citation info and the PDF. I’ll let you know if we have any questions, and if others have similar issues, please weigh in!

4 years ago
Changed the status to
Under Consideration
4 years ago

We had this happen with another nest a couple weeks ago, where the incorrect pdf (from another study in the nest) was uploaded and tagged, and caught during QC due to being a duplicate of a study the person had already QC’d. Obviously the initial issue is double checking that the chosen pdf matches the source when uploading, but I think another factor that prolonged the issue is that the full text page doesn’t include any of NK’s bibliographic data, so someone would have had to open the study directly from study inspector or flip to the abstract tab to realize that the full text didn’t match the citation. A warning for title matching would be helpful, and maybe another option would be putting the author/year at the top with the pdf tools so there’s some matching information presented with the full text for confirmation!

4 years ago

Thanks for the anecdote! I’m slightly hesitant to tie title detection/matching to FT upload because:

  • It can be inaccurate (guessing ~1% of the time IME), and false positives may prove annoying
  • The software we use for extracting bibliographic data from PDFs is mildly slow; delays in upload could prove frustrating especially if you’re trying to do so in fast succession

Both of these problems may be avoided with better positioning (e.g. a red warning icon at the top of the viewer, that displays only after processing has finished & a mismatch may be present). I’m fond of your simple recommendation to display author/year more plainly as well! @Jeffrey Johnson do you have time to test designs for these ideas?

4 years ago
Changed the status to
Archived
6 days ago