Skip to main content

graphical and continuous representations of ICD-9 and ICD-10 codes

Project description

PyPI version fury.io Documentation Status Downloads DOI

Thank you for your interest in ICD codex! I (Jeremy) wrote this in 2020 as part of class project and it has gotten quite a few downloads. However, since then, sequence representation has significantly improved. At this point, **I would not recommend using node2vec to represent ICD codes.** Instead, use a [language model](https://platform.openai.com/docs/guides/embeddings). The node2vec functionality is provided for compatibility with existing projects.

The `networkx` hierarchy should still remains useful.

If there is interest in extending this library for use with modern sequence learning algorithms, please reach out.

What is it?

A python library for building vector representations of ICD-9 and ICD-10 codes. (2025 comment: the vector representations here are constructed using outdated algorithms.) Because it takes advantage of the hierarchical nature of ICD codes, it also provides these hierarchies in a networkx format. (2025 comment: this data structure should still remain useful.)

Motivation

icdcodex was the first prize winner in the Data Driven Healthcare Track of John Hopkins' MedHacks 2020. It was hacked together to address the problem of ICD miscodes, which is a major issue for health insurance in the United States. Indeed, while ICD coding is tedious and labour intensive, it is not obvious how to automate because the output space is enourmous. For example, ICD-10 CM (clinical modification) has over 70,000 codes and growing.

There are many strategies for target encoding that address these issues. icdcodex has two features that make ICD classification more amenable to modeling:

  • Access to a networkx tree representation of the ICD-9 and ICD-10 hierarchies
  • Vector embeddings of ICD codes using the node2vec algorithm (including pre-computed embeddings and an interface to create new embeddings)

Example Code

from icdcodex import icd2vec, hierarchy
embedder = icd2vec.Icd2Vec(num_embedding_dimensions=64)
embedder.fit(*hierarchy.icd9())
X = get_patient_covariates()
y = embedder.to_vec(["0010"])  # Cholera due to vibrio cholerae

In this case, y is a 64-dimensional vector close to other Infectious And Parasitic Diseases codes.

Related Work

The Hackathon Team

  • Jeremy Fisher (Maintainer)
  • Alhusain Abdalla
  • Natasha Nehra
  • Tejas Patel
  • Hamrish Saravanakumar

Documentation

See the full documentation: https://icd-codex.readthedocs.io/en/latest/

Contributions

Contributions are always welcome!

History

0.5.2 (2025-01-05)

  • Added 2025 ICD-10-CM
  • Bumped deps
  • Removed EOL python's from CI
  • Fix bug where 2024 codes are being pulled from 2023
  • Fixed some flake8 bugs
  • Fixed use of deprecated field in networkx

0.4.9, 0.5.0 and 0.5.1 (2024-01-08)

  • Added 2024 ICD-10-CM
  • Fixed a bug where the node descriptions were being discarded
  • Regenerated the hierarchy files with the descriptions

0.4.8 (2023-05-02)

  • Use the newer scikit-learn PyPI location
  • Deprecate the Python 3.6 version
  • Add 2023 ICD-10-CM

0.4.6 (2022-10-13)

  • Add 2022 ICD-10-CM (thank you @keithcallenberg!)

0.4.4 and 0.4.5 (2020-10-18)

  • Add the code descriptions for ICD9
  • Add usage on how to recapitulate functionality of sirrice/icd9
  • Make the hierarchy directed to allow simpler and more intuitive traversal
  • Fix issue where edges were not being formed between "Diseases Of The Blood And Blood-Forming Organs" and "Congenital Anomalies" and their children

0.4.3 (2020-10-04)

  • Fix issue where hierarchy jsons were not being shipped with the pypi distribution

0.4.2 (2020-10-03)

  • Add support for python <= 3.8 in the hierarchy module by using the importlib.resources backport

0.4.1 (2020-09-11)

  • Update PyPI metadata

0.4.0 (2020-09-11)

  • ICD-10-CM (2019 to 2020) codes are now fully present (whereas hackathon version missed certain codes)
  • Versions of the ICD 9 and ICD-10-CM hierarchies are now cached to the data module
  • Changed the hierarchy API: hierarchy.icd9hierarchy() is now hierarchy.icd9(). Ditto for ICD-10-CM.

0.3.0 (2020-09-05)

  • Finesse API, now consistent between documentation and implementation

0.1.0 (2020-09-04)

  • First release on PyPI, testing the waters during hackathon

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

icdcodex-0.5.2.tar.gz (29.7 MB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

icdcodex-0.5.2-py2.py3-none-any.whl (6.7 MB view details)

Uploaded Python 2Python 3

File details

Details for the file icdcodex-0.5.2.tar.gz.

File metadata

  • Download URL: icdcodex-0.5.2.tar.gz
  • Upload date:
  • Size: 29.7 MB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/4.0.2 CPython/3.11.4

File hashes

Hashes for icdcodex-0.5.2.tar.gz
Algorithm Hash digest
SHA256 5930ff8c04d1d4e970bfce87927ad03e6a19d078c949753ba4b9defbe3522171
MD5 c7c9aa51c0f844e064302050d0253468
BLAKE2b-256 724d250db9de37f4b553132c87fb4b1e68c3989ed6369227679feac0c92403eb

See more details on using hashes here.

File details

Details for the file icdcodex-0.5.2-py2.py3-none-any.whl.

File metadata

  • Download URL: icdcodex-0.5.2-py2.py3-none-any.whl
  • Upload date:
  • Size: 6.7 MB
  • Tags: Python 2, Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/4.0.2 CPython/3.11.4

File hashes

Hashes for icdcodex-0.5.2-py2.py3-none-any.whl
Algorithm Hash digest
SHA256 db8a7c3c3902fe24508b65711ea3f38f08dc185c38ff0b66e213c60ad82a4e6b
MD5 c545cdbce710f4b7603b67a40a84e352
BLAKE2b-256 ae764eb30cdd85c50bf54a6931cddcb6c1136f53ef2669baba3c32325bb220d7

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page