joniponi/multilabel_inpatient_comments_16labels

HCAHPS survey comments multilabel classification

This model is a fine-tuned version of Bio_ClinicalBERT on a dataset of HCAHPS survey comments.

It achieves the following results on the evaluation set:

precision recall f1-score support

medical 0.87 0.81 0.84 83 environmental 0.77 0.91 0.84 93 administration 0.58 0.32 0.41 22 communication 0.85 0.82 0.84 50 condition 0.42 0.52 0.46 29 treatment 0.90 0.78 0.83 68 food 0.92 0.94 0.93 36 clean 0.65 0.83 0.73 18 bathroom 0.64 0.64 0.64 14 discharge 0.83 0.83 0.83 24 wait 0.96 1.00 0.98 24 financial 0.44 1.00 0.62 4 extra_nice 0.20 0.13 0.16 23 rude 1.00 0.64 0.78 11 nurse 0.92 0.98 0.95 110 doctor 0.96 0.84 0.90 57

micro avg 0.81 0.81 0.81 666 macro avg 0.75 0.75 0.73 666 weighted avg 0.82 0.81 0.81 666 samples avg 0.64 0.64 0.62 666

Model description

The model classifies free-text comments into the following labels

• —Medical
• —Environmental
• —Administration
• —Communication
• —Condition
• —Treatment
• —Food
• —Clean
• —Bathroom
• —Discharge
• —Wait
• —Financial
• —Extra_nice
• —Rude
• —Nurse
• —Doctor

How to use

You can now use the models directly through the transformers library. Check out the model's page for instructions on how to use the models within the Transformers library.

Load the model via the transformers library:

from transformers import AutoTokenizer, AutoModel
tokenizer = AutoTokenizer.from_pretrained("joniponi/multilabel_inpatient_comments_16labels")
model = AutoModel.from_pretrained("joniponi/multilabel_inpatient_comments_16labels")