Name Entity Recognition for Medical Transcripts

Performing NER on MEdical Datasets

Running the Project

Creating the Environment

1. Install Pytorch

   Pytorch is a requirement for running some of the libraries. However, pytorch implementation varies according to machine, so it has to be done manually.

   Refer to https://pytorch.org/get-started/locally/ for installation.

2. Installing requirements

   Requirements can be installed using pipenv dependency manager. Install pipenv using command:

   pip install pipenv

   Then you can install the requirements using the command:

   pipenv install

   Note: The dependencies are a bit large so a stable internet connection and a bit of patience is required

3. Installing language model

   We need to download a spacy model. We can download it by the command:

   python -m spacy download en_core_web_lg

Running the Notebook

1. We need jupyter installed to run the notebook

   pip install jupyter

2. Launch the notebook interface using command

   jupyter notebook

3. Navigate to the required notebook to view

Running From CLI

We can run the NER data extractor from the cli too by using the command from the base dir

python ner/medical_ner.py <location-to-file>

For instance,

python ner/medical_ner.py assets/txt_reports/report_0.txt