Facilitating phenotyping from clinical texts: the medkit library.

Area of Science:

• Biomedical Informatics
• Computational Health Sciences
• Health Data Science

Background:

• Phenotyping using Electronic Health Records (EHRs) is vital for clinical research.
• Extracting information from unstructured clinical text presents significant challenges due to heterogeneity and complexity.
• These challenges lead to time and cost constraints in observational studies.

Purpose of the Study:

• To develop an open-source Python library to facilitate the development, evaluation, and reproducibility of phenotyping pipelines.
• To address the tedious nature of phenotyping from clinical text.
• To support the secondary use of EHRs in research.

Main Methods:

• Developed medkit, an open-source Python library.
• medkit enables the composition of data processing pipelines using reusable software bricks called medkit operations.
• Shared pre-developed operations and pipelines for community use and enrichment.

Main Results:

• medkit provides a framework for building and evaluating phenotyping pipelines.
• The library promotes code reuse and standardization in phenotyping workflows.
• Facilitates easier development and validation of phenotyping algorithms.

Conclusions:

• medkit simplifies and accelerates the process of phenotyping from clinical text.
• The open-source library enhances the reproducibility and efficiency of EHR-based research.
• Encourages community collaboration for advancing phenotyping methodologies.