Skip to main content
ARS Home » Southeast Area » Little Rock, Arkansas » Microbiome and Metabolism Research Unit » Microbiome and Metabolism Research » Research » Publications at this Location » Publication #436579
Research Project: Early Life Factors and Microbiota Impact on Healthy Development

Location: Microbiome and Metabolism Research

Title: Modeling Continuous 24-Hour Ambulatory Blood Pressure Trajectories Using Penalized Spline Mixed-Effects Regression

Author
item CHUNG, SIMON - University Arkansas For Medical Sciences (UAMS)
item ADAMS, SEAN - University Of California, Davis
item BLASZAK, RICHARD - University Arkansas For Medical Sciences (UAMS)
item BORSHEIM, ELISABET - University Arkansas For Medical Sciences (UAMS)
item MOHAMMAD, ILYAS - University Arkansas For Medical Sciences (UAMS)
item POTTS, CHRISTINA - Arkansas Children'S Hospital
item RYAN, BRANDI - Arkansas Children'S Hospital
item RIVERA, JENY - University Arkansas For Medical Sciences (UAMS)
item DIAZ, EVA - University Arkansas For Medical Sciences (UAMS)
item BROWN, ANDREW BROWN - University Arkansas For Medical Sciences (UAMS)

Submitted to: Meeting Abstract
Publication Type: Abstract Only
Publication Acceptance Date: 4/6/2026
Publication Date: N/A
Citation: N/A

Interpretive Summary:

Technical Abstract: BACKGROUND: 24-hour ambulatory blood pressure monitoring (ABPM) provides dense, repeated measures essential for understanding cardiometabolic health. The complex multi-level data structure poses analytical challenges for quantifying effects of interventions on time-varying outcomes. Traditional statistical approaches summarize measurements into crude daytime and nighttime averages, which discard valuable temporal dynamics and fails to capture potential non-linear blood pressure changes. OBJECTIVE: To model continuous ABPM profiles after exercise versus no exercise in adolescents with excess weight. METHODS: We analyzed 24-hour ABPM data (Systolic, Diastolic, Mean Arterial, and Pulse Pressures, and Heart Rate) from an ongoing trial testing acute effects of isometric versus aerobic exercises. Because intervention groups remain blinded, the present analysis compares ABPM dynamics 24 hours following a no-exercise baseline versus exercise, pooled across exercise condition groups using mixed-effects generalized additive models (GAMs). The use of penalized cubic splines relaxes assumptions on the relationship between ABPM measurements and time. Datapoints were aligned using time-of-day, time-at-sleep and time-after-exercise; analysis was repeated under different alignments. Instead of fitting isolated trajectories, we modeled differences by estimating global reference curves for the no-exercise baseline and curves representing the physiological change following exercise. Subject-specific random intercepts were included to account for correlation within individuals. We tested for statistically significant temporal shape deviations in the trajectories. RESULTS: 38 adolescents provided 76 ABPM profiles. GAMs successfully parameterized both the overall mean shifts and the non-linear temporal differences in blood pressure before and after exercise. Post-exercise difference smooths significantly deviated from zero (all p<0.05) across all metrics and alignments, excepting diastolic under time-of-day alignment. CONCLUSIONS: Modeling ABPM data as continuous processes with mixed-effects GAMs rather than summarizing within time periods using means provides time-sensitive understanding of blood pressure fluctuations across 24 hours. Future analyses will generalize this approach for time-by-exercise modality interactions in ABPM dynamics. FUNDING: NIH-NIGMS P20GM109096. USDA-ARS 6026-10700-001-000D