Exploring long-term breast cancer survivors' care trajectories using dynamic time warping-based unsupervised...

Area of Science:

• Oncology
• Health Services Research
• Data Science

Background:

• Long-term breast cancer survivors (BCS) require dedicated clinical follow-up due to rising numbers and complex care needs.
• Understanding the healthcare utilization patterns of BCS is crucial for optimizing survivorship care.

Purpose of the Study:

• To identify and visualize temporal patterns in the healthcare trajectories of long-term breast cancer survivors.
• To compare healthcare service utilization between BCS and a control group without breast cancer.

Main Methods:

• A dynamic time warping-based unsupervised clustering methodology was applied to the care trajectories of 6214 female BCS.
• Care-transition patterns were extracted and visualized using directed network diagrams.
• A control group of 12,412 females without breast cancer was used for comparison.

Main Results:

• Breast cancer survivors exhibit more intense and complex use of healthcare services, including radiology, outpatient care, and hospital admissions.
• Higher mortality rates and increased comorbidities were observed in various care transitions for BCS compared to controls.
• Individual service transitions revealed significant patient and temporal information.

Conclusions:

• The methodology effectively identifies and visualizes hidden temporal patterns in BCS healthcare service usage.
• Findings can inform better understanding of BCS health system navigation, enabling more accurate prediction of needs and personalized care plans.