|
|
--- |
|
|
license: apache-2.0 |
|
|
datasets: |
|
|
- ARI-HIPA-AI-Team/Dataset |
|
|
language: |
|
|
- en |
|
|
tags: |
|
|
- healthcare |
|
|
- HIPAA |
|
|
- federal_regulations |
|
|
pipeline_tag: text-classification |
|
|
library_name: keras |
|
|
--- |
|
|
|
|
|
# HIPA-AI ML Model: Keras |
|
|
This model takes paragraph style reddit posts and predicts whether they are a violation of the Health Insurance Portability and Accountability Act. |
|
|
|
|
|
## Source |
|
|
This model was the first place finisher in the HIPA-AI Model competition hosted by Codalab with a score of 70% accuracy. It was authored by tarak2134. |
|
|
|
|
|
## Details |
|
|
This model takes a text body between 40 and 750 words as input. |
|
|
As output, the model gives a binary; "yes, this is a violation", or "no, this is not a violation" |
|
|
The model tends to be more accurate on larger bodies of text. |
|
|
|
|
|
## Dataset |
|
|
This model was trained on an annotated set of 100 reddit posts. To get more information, see the dataset card at ARI-HIPA-AI-Team/Dataset |
|
|
|
|
|
## Model Card Authors |
|
|
Brayden Cloutier |
|
|
|
|
|
## Model Card Contact |
|
|
brayclou@umich.edu, zmorell@umich.edu |
