I’m pleased to say that my friend and colleague Gandhi Rajan and I are noting that our REST layer over the popular cTAKES NLP project** is ready. Perhaps this could be useful to some here! I’d love to hear about any potential use cases folks have in mind.
** Apache cTAKES ™ is a natural language processing system for extraction of information from electronic medical record clinical free-text.
(Please let me know if you are a MacOS or Windows user and would like to help document the install steps - currently, they are Linux oriented)
This is very interesting - thanks @MatthewVita for posting. I’ve been thinking about new ways of data input for clinical interfaces recently, and I hadn’t realised there was an Apache project in this area.
A while ago I was working on a simple spec for a kind of ‘medical markdown’ to help you annotate free text and enable it to be regularly parsed into structured text, so that clinicians wouldn’t need to constantly mouse between text boxes in healthcare applications. https://github.com/open-health-hub/medical-markdown I think I ran out of ideas with this but would welcome some other clinician inputs or tech expertise on implementation.
Brilliant - I’ve looked at cTakes before for a variety of projects that unfortunately didn’t get off the ground. One such project was to look for patients that had been prescribed a drug that subsequently had alerts issued. For instance, Esmya -ullipristal acetate for fibroids - now known to cause liver problems. The prescription info can either be in the primary care system, or if it was prescribed by secondary care in a setting without HEPMA, it might be mentioned in clinic letters. I thought about using cTakes to NLP our digital letter repository to search for the drug and return a list of patents to whom it may have been prescribed.
That’s one example, but I have a few others.