An open-source framework for end-to-end analysis of electronic health record data

Area of Science:

• Computational biology
• Health informatics
• Epidemiology

Background:

• Digitalization has led to widespread electronic health record (EHR) data collection.
• Existing frameworks lack extensibility for heterogeneous EHR data analysis.
• A standardized approach is needed for comprehensive exploratory analysis of diverse health data.

Purpose of the Study:

• Introduce ehrapy, a modular, open-source Python framework for EHR data analysis.
• Provide a comprehensive toolkit for exploratory analysis of heterogeneous epidemiological and EHR data.
• Facilitate data sharing and deep learning model training using ontologies.

Main Methods:

• ehrapy integrates data extraction, quality control, and low-dimensional representation generation.
• Includes statistical modules for patient stratification, differential comparison, survival analysis, and causal inference.
• Leverages ontologies for data standardization and interoperability.

Main Results:

• Demonstrated ehrapy's utility across six diverse case studies.
• Stratified pneumonia patients into finer phenotypes and identified survival biomarkers.
• Quantified medication effects on length of stay and analyzed cardiovascular risks.
• Reconstructed SARS-CoV-2 disease trajectories and detected/mitigated EHR data biases.

Conclusions:

• ehrapy offers a standardized framework for EHR data analysis.
• It serves as a foundational tool for biomedical research and community-driven analysis.
• Enables advanced analyses including deep learning model development on EHR data.