Multivariate Time-Series Forecasting of 24-Hour Ambulatory Blood Pressure Using Long Short-Term Memory and Temporal Fusion Transformers

Author

• Sebastian Alejos Torres, The University of Texas Rio Grande Valley

Date of Award

5-1-2026

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

Time series play a central role in healthcare by enabling continuous patient monitoring and forecasting of physiological and clinical measurements. Traditional models, such as autoregressive integrated moving average (ARIMA), are limited in capturing nonlinear dynamics and irregular sampling. In this study, we develop and evaluate Long Short-Term Memory (LSTM) networks and Temporal Fusion Transformers (TFTs) to forecast 24-hour ambulatory systolic and diastolic blood pressure (SBP and DBP) time series enriched with demographic and clinical features. Mean absolute error (MAE), root mean squared error (RMSE), mean absolute percentage error (MAPE), and R² were used to evaluate predictive accuracy and temporal pattern learning. Results demonstrate that LSTM consistently outperforms TFT across most configurations. To our knowledge, this is the first study to apply TFTs to 24-hour ambulatory blood pressure (ABP) recordings and the first to directly compare TFT and LSTM models in this context, highlighting their potential for clinical monitoring and cardiovascular risk assessment.

Comments

Copyright 2026 Sebastian Alejos Torres. All Rights Reserved. https://proquest.com/docview/3371452247

Recommended Citation

Torres, S. A. (2026). Multivariate Time-Series Forecasting of 24-Hour Ambulatory Blood Pressure Using Long Short-Term Memory and Temporal Fusion Transformers [Master's thesis, The University of Texas Rio Grande Valley]. ScholarWorks @ UTRGV. https://scholarworks.utrgv.edu/etd/1966