aura-healthcare / seizure_detection_pipeline Goto Github PK

There are some places where the script fetch_database.py is hardcoded to use the teppe format instead of tuh. We can fix this locally for now by adding the following to the script:

TEPPE_PATIENT_PATTERN = TUH_PATIENT_PATTERN
DB = 'tuh'

but that is just a temporary fix. It seems like the script was intended to support both formats, so there should be a better way of fixing this problem.

I have followed the steps described in the repo's readme to setup the Aura docker images.

After issuing the docker-compose up command, images are being built and after some time setup is ready. But when I try to access:

• port 80 it redirects to https, no connection.
• port 8082 I get a 403 forbidden
• port 5000 (mlflow) the list of experiments/models is empty.
• port 8080 (airflow) I get a "DAG import error" saying "ModuleNotFoundError: No module named 'ecg_qc'"

Full error report in the 8080 UI is

Broken DAG: [/opt/airflow/dags/seizure_detection_pipeline.py] Traceback (most recent call last):
  File "/opt/airflow/dags/seizure_detection_pipeline.py", line 7, in <module>
    from src.usecase.apply_ecg_qc import apply_ecg_qc
  File "/opt/airflow/src/usecase/apply_ecg_qc.py", line 5, in <module>
    import ecg_qc
ModuleNotFoundError: No module named 'ecg_qc'

Docker log is attached below:
aura_setup.log