Welcome to the Emory Health AI Bias Datathon!
You can find the summary slides from each team's presentations at this link
Overview:
Please find your server assignments by team below. The schedule for the weekend is available on our website.
.yml files have been provided to set up GPU accelerated environments for either Tensorflow or Pytorch. You can test whether your environment is correctly configured with:
To install the environment and add the kernel to Jupyterhub
conda env create -f <env-config-file>.yml
conda activate <env-name>
python -m ipykernel install --name=<env-name> --user
After running the above commands refresh your browser window.
Note: Without the --user flag this command will fail
Tensorflow:
python -c "import tensorflow as tf; print(tf.config.list_physical_devices('GPU'))"
PyTorch:
python -c "import torch; print(torch.cuda.is_available();"
Server/Room Assignments:
| Team | Dataset | Server | Room | Project Statement |
|---|---|---|---|---|
| 1 + 9 | MRKR #1 + #2 | jhub9.datathon.org, jhub8.datathon.org | 188 | Validating the ALP score for Osteoarthritis assessment. |
| 2 | EMBED #1 | jhub.datathon.org | 251 | Detecting and mitigating bias in medical imaging algorithms. |
| 3 | EMBED #2 | jhub2.datathon.org | 253 | Detecting and mitigating bias in medical imaging algorithms. |
| 4 + 11 | Various Datasets #1 + #2 | jhub4.datathon.org, jhub10.datathon.org | 325 | Are new forms of dataset good enough for AI model development ? What are their impacts on bias? |
| 5 + 8 + 12 | ICU #1 + #2 + #3 | jhub1.datathon.org, jhub7.datathon.org, jhub11.datathon.org | 353 | Evaluate biases in the Emory ICU dataset and disparities proxies. |
| 6 | CXR #1 | jhub5.datathon.org | 269 | Detecting and mitigating bias in medical imaging algorithms. |
| 7 | CXR #2 | jhub6.datathon.org | 268 | Detecting and mitigating bias in medical imaging algorithms. |
| 10 | EMBED #3 | jhub3.datathon.org | 255 | Detecting and mitigating bias in medical imaging algorithms. |
| 13 | ChatGPT | jhub12.datathon.org | 130 | Red Teaming to test the performance of large language models (ChatGPT for health). |
Room Assignment Maps:
