Clustering 24-Hour Ambulatory Blood Pressure Time Series With Dynamic Time Warping and Time Warp Edit Distance

Author

John Knight, The University of Texas Rio Grande Valley

Date of Award

12-1-2025

Document Type

Thesis

Degree Name

Master of Science (MS)

Department

Applied Statistics and Data Science

First Advisor

Kristina Vatcheva

Second Advisor

Yifeng Gao

Third Advisor

Thoa Thieu

Abstract

Ambulatory blood pressure monitoring (ABPM) captures dynamic circadian changes in blood pressure (BP) that are not reflected in static clinic readings. This study applied time-series clustering with two elastic distance measures—Dynamic Time Warping (DTW) and Time Warp Edit Distance (TWED)—to identify distinct phenotypes in 2,155 24-hour ABPM time series from participants in the Maracaibo Aging Study. DTW and TWED both yielded three clusters corresponding to non-dipping, moderate-dipping, and strong-dipping patterns. The non-dipping group showed elevated nighttime BP, associated with greater cardiovascular and cognitive risk, while the strong-dipping group was associated with higher education and younger age in baseline clinic measurements. TWED indicated greater variability in non-dippers, whereas DTW results were more mixed. The TWED method was preferred overall for its clinically interpretable clusters and greater flexibility in handling irregular data. Results demonstrated that unsupervised time-series methods can capture clinically meaningful BP phenotypes beyond conventional summary metrics, supporting the use of elastic distance clustering for precision risk stratification in hypertension and aging research.

Comments

Copyright 2025 John Knight. All Rights Reserved. https://proquest.com/docview/3292558708

Recommended Citation

Knight, J. (2025). Clustering 24-Hour Ambulatory Blood Pressure Time Series With Dynamic Time Warping and Time Warp Edit Distance [Master's thesis, The University of Texas Rio Grande Valley]. ScholarWorks @ UTRGV. https://scholarworks.utrgv.edu/etd/1817