pytorch expert for prediction of acute respiratory disease using llama on clinical notes
Budget: $30 – $50 USD
Code containing pre-trained llama model that will be able to predict acute respiratory failure based on all clinical notes present in mimic-iv-note.
In the clinical note db there is a part called "Discharge diagnoses" that slightly change from patient to patient. Normally that part is masked/unknow so model needs to be tested when that is not present.
The code needs to work even if the structure of the text column changes.
Common classification evaluation metric needs to be included.
The code needs to be able to be concatenated with a image classification model (Densenet121) in the future and compared to that.
If training time takes too long, do it with a smaller db and I will do it myself once the whole code is ready.
No plagiarism.
Possibly using pytorch.
Also able to test model locally with a small example even if the outcome is not really accurate in the test.
Clear comments and explanations for each step.
database is mimic-note-iv that I downloaded here: https://drive.google.com/drive/folders/1Z5a8mu_8wq4fH6awyATSWFBVVcwsHyhJ?usp=sharing
In the clinical note db there is a part called "Discharge diagnoses" that slightly change from patient to patient. Normally that part is masked/unknow so model needs to be tested when that is not present.
The code needs to work even if the structure of the text column changes.
Common classification evaluation metric needs to be included.
The code needs to be able to be concatenated with a image classification model (Densenet121) in the future and compared to that.
If training time takes too long, do it with a smaller db and I will do it myself once the whole code is ready.
No plagiarism.
Possibly using pytorch.
Also able to test model locally with a small example even if the outcome is not really accurate in the test.
Clear comments and explanations for each step.
database is mimic-note-iv that I downloaded here: https://drive.google.com/drive/folders/1Z5a8mu_8wq4fH6awyATSWFBVVcwsHyhJ?usp=sharing