code for modular summarization work published in ACL2021 by Krishna et al

This repository contains the code for running modular summarization pipelines as described in the publication
Krishna K, Khosla K, Bigham J, Lipton ZC. Generating SOAP Notes from Doctor-Patient Conversations." ACL 2021.

Instructions

Although we can not release models trained on the confidential medical data, we have released models trained on the publicly available AMI dataset.
To reproduce the results on the AMI dataset, you need to follow the steps listed below. For convenience, we have also created a Google Colab notebook here that runs these steps on Google's servers (free-of-cost as of June 2021) and produces the summaries and their rouge scores.

Step1: Set up the environment by installing the required packages mentioned in requirements.txt using pip.

Step2: Download the ami_models folder from this link and put it at the root of the repository:

Step3: Run the following 3 commands to prepare data, run summary generation pipelines, and show the achieved rouge scores.

# command1: downloads and preprocesses AMI dataset  
./prepare_data.sh  
  
 # command2: runs the summarization pipelines on the data and computes rouge scores  
 # (before running this command, you need to download the models as shown above)  
./predict_ami.sh  
  
# command3: print the results  
python show_results.py