Paving the COWpath: Learning and visualizing clinical pathways from electronic health record data.

Area of Science:

• Health Informatics
• Clinical Data Science
• Translational Medicine

Background:

• Clinical pathways are essential for translating evidence into practice.
• Developing patient-centered pathways aligned with clinical workflows is crucial.
• Electronic Health Records (EHR) offer rich data for pathway extraction.

Purpose of the Study:

• To propose a practice-based clinical pathway development process.
• To introduce a data-driven methodology for extracting common clinical pathways from EHR data.
• To create patient-centered pathways consistent with clinical workflow and evidence-based care.

Main Methods:

• Extracted visit data for 1,576 chronic kidney disease (CKD) patients who developed acute kidney injury (AKI) from 2009-2013.
• Modeled multi-dimensional clinical records into one-dimensional sequences capturing visit purpose, procedures, medications, and diagnoses.
• Analyzed and clustered visit sequences, characterizing them as Markov chains to extract and visualize significant transitions into clinical pathways.

Main Results:

• Identified 31 distinct patient subgroups with extracted clinical pathways.
• Gained insights into patient condition progression and medication prescription patterns over time.
• Visualized pathways depicting disease progression likelihood and direction under varied contexts, highlighting guideline-consistent practices and improvements.

Conclusions:

• EHR data accuracy and patient diversity present challenges in learning insightful pathways.
• Learning and visualizing pathways from EHR data can enhance practice review for healthcare providers.
• This approach supports patient engagement in shared decision-making regarding their care.