00:01
Where in the world is hematology?
~5000 abstracts reviewed, ~500 had author from LMICs13
Where in the world is hematology?
manual extraction is slow and ruins your back4
Right [patient, drug, time, dose, route, place]
Léon Tshilolo, MD, PhD. ASH 2018, speaking on hydroxyurea in Sub-Saharan Africa.5
Myopia
6
Gatekeeper/Guide
7
Current project
What if we could automate the extraction and structuring of all ~100k abstracts?9
Current project
keep it public, keep it safe10
Current project
analysis pipeline11
Current project
BYOA (bring your own analysis)12
Planned analyses
- trends
- geographic
- chronologic
- topologic
- ontologic
- biographic
- disparities
- e.g. overlay maps of the disease itself (incidence, morbidity, etc.) and research about that disease
- funding/resource allocation
h/t Hans Rosling13
Challenges
Structure, standards
14
Challenges
missing, incomplete, wrong16
Challenges
Web scraping is the bug. Not sure who the other guy is.17
Challenges (also a feature)
If you can write SQL, you can analyze the data.18
Challenges (also a feature)
ChatGPT to the rescue?19
