Usage

Installation

To use PyUMLS-Similarity, first install it using pip:

(.venv) $ pip install PyUMLS-Similarity

Computing Semantic Similarity

PyUMLS-Similarity allows you to compute the semantic similarity between medical concepts from the UMLS database. To start, import the package and initialize it with your UMLS MySQL database information:

from pyumls_similarity import PyUMLS_Similarity
mysql_info = {
    "username": "your_username",
    "password": "your_password",
    "hostname": "localhost",
    "socket": "your_socket",
    "database": "umls"
}
umls_sim = PyUMLS_Similarity(mysql_info)

You can now use the umls_sim.similarity() function to compute similarity between pairs of concepts:

cui_pairs = [
    ('C0018563', 'C0037303'),  # Example CUI pairs
    ('C0035078', 'C0035078'),
    # ... more pairs
]
measures = ['lch', 'wup']
similarity_df = umls_sim.similarity(cui_pairs, measures)

Handling Exceptions

The PyUMLS-Similarity package may raise exceptions if it encounters invalid input or other errors. For example, if an invalid CUI pair is provided, the package will raise a InvalidCUIPairError:

For example:

>>> umls_sim.similarity([('INVALID_CUI', 'C0037303')], measures)
Traceback (most recent call last):
  ...
pyumls_similarity.InvalidCUIPairError: Invalid CUI pair provided