On Learning and Visualizing Practice-based Clinical Pathways for Chronic Kidney Disease.

Area of Science:

• Nephrology
• Health Services Research
• Data Science

Background:

• Chronic Kidney Disease (CKD) impacts millions of US adults, presenting significant clinical and economic challenges.
• Care delivery models are being explored for potential improvements in patient outcomes, experience, and cost-effectiveness.
• Understanding patient journeys is crucial for developing effective, personalized treatment strategies.

Purpose of the Study:

• To analyze real-world CKD patient data to identify distinct clinical pathways.
• To discover patient subgroups based on visit history and associated health conditions.
• To model common treatment trajectories and durations within identified patient groups.

Main Methods:

• Utilized hierarchical clustering based on Longest Common Subsequence (LCS) distance.
• Modeled patient visit history as sequences of visit types, dates, procedures, and diagnoses.
• Analyzed 8,553 CKD patients' treatment data to identify six distinct patient subgroups.

Main Results:

• Discovered six patient subgroups with significant differences in demographics and health conditions.
• Elicited high-probability visit transitions within each subgroup to define common clinical pathways.
• Characterized distinct treatment durations and patterns for each identified patient group.

Conclusions:

• The study provides evidence for patient-centered treatment approaches in Chronic Kidney Disease (CKD).
• Identified clinical pathways can empower patients in managing their disease and complications.
• Findings offer a review guide for clinicians to optimize CKD care delivery.