Dataset used for training
Would you share that dataset(s) you used for model training? Is there a paper associated with this work?
Thanks in advance.
Appreciate the interest. Checked this one a bit late. Unfortunately, I can't provide it at the moment. i have not written a paper. It is relatively self explanatory. Entailment are cases where the condition is present with the patient. Contradiction is where a patient does not have the issue. If you are curious on any particular condition or if something needs to be improved upon, please let me know.
Did you train this on clinical notes - ie. patient notes from the electronic medical record or some other source?
This was trained on synthetic data sets. But it should scale to clinical notes well. Let me know if you run into any issues.
Is it possible to share the dataset at all? even the dummy schema would do @reachosen
@vchitturi please give a look to the helper harness located at https://github.com/reachosen/SDOHv7. Should be of guidance. Feel free to email me [email protected] for any questions.