Recognizes diagnoses in histopathology reports #13388
Unanswered
marcottemick
asked this question in
Help: Coding & Implementations
Replies: 0 comments
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
-
Hello, I would like to train a model to recognize diagnoses in anatomical pathology operating accounts in French. The reports were generated by an LLM Mistral IA. The texts were generated from 20 examples of diagnoses. I currently have 1600 operating reports. Here is an example of an operating report in which I want to recover 'hémangiome'.
I tried both possibilities offered by spacy to generate a ner + tok2vec training pipeline but the results are always the same. Example :
DIAG P : 87.50 R : 9.30 F : 16.82
Whereas in training I obtain a score of 0.98 and he quickly obtains this score with few epochs.
The 20 diagnostic categories are balanced and I tried to do data augmentation to reach 2000 reports per diagnosis. I also try to play with the hyperparameters, but that doesn't change anything either.
I thank in advance any answers that will be given.
(text translated with Google trad from French)
Beta Was this translation helpful? Give feedback.
All reactions