Hemodynamic Fidelity Across Anatomical Acquisition Sites: Stationary Blood Pressure Concordance and Dynamic Treadmill Heart Rate Tracking of a Plantar Multi-Sensor Platform Versus Finger and Wrist Photoplethysmography in 483 Adults

ABSTRACTBackground: Resting agreement between wearable sensors and clinical reference standards does not establish fidelity during dynamic cardiovascular exertion, nor do software-imported blood pressure logs represent direct hemodynamic sensing. This study evaluated stationary blood pressure (BP) and heart rate (HR) agreement of a fixed ground-contact plantar platform alongside resting and dynamic treadmill HR tracking of finger-worn (OURA) and volar wrist-worn (WHOOP) photoplethysmography (PPG) devices in an established user cohort. Methods: In 483 adults (267 women, 216 men) recruited from active personal owners with mature device baselines, the Deep Health® plantar platform (model PH-CS20C3) was compared concurrently against manual brachial auscultation and 60-second radial pulse counts during quiet stance. OURA and WHOOP HR outputs were evaluated at rest and across a 30-minute treadmill protocol (8:00 min/mile [7.5 mph], 2.0° incline) sampled at baseline and every 3 minutes (5,313 paired observations per wearable). Agreement was assessed via mean bias (MD), standard deviation of differences (SDD), Bland–Altman 95% limits of agreement (LoA), mean absolute error (MAE), root mean square error (RMSE), Lin’s concordance correlation coefficient (CCC), and 10,000-iteration participant-level bootstrap confidence intervals. Wearable BP entries were excluded from primary accuracy testing because audit confirmed they reflected static external cuff imports or uncalibrated overnight estimates rather than contemporaneous sensor measurements. Results: Stationary plantar SBP and DBP biases were both −0.01 mmHg, with SDDs of 1.09 mmHg (95% LoA: −2.15 to +2.13 mmHg; CCC = 0.9890) and 0.75 mmHg (95% LoA: −1.48 to +1.45 mmHg; CCC = 0.9932), respectively, with 93.8% of SBP and 99.6% of DBP readings falling within ±2.0 mmHg of manual auscultation. Stationary plantar HR bias was +0.04 bpm (MAE = 0.34 bpm), whereas stationary OURA and WHOOP HR MAEs were 8.50 and 13.85 bpm. During treadmill exercise, reference HR surged by +60.38 bpm at minute 3 (145.51 bpm), whereas OURA and WHOOP lagged at 134.04 bpm (MD = −11.47 bpm) and 135.23 bpm (MD = −10.28 bpm). Across minutes 3–30, overall exercise MAE was 8.43 bpm (95% CI: 8.27–8.59) for OURA and 7.66 bpm (95% CI: 7.50–7.81) for WHOOP. During late exercise (minutes 12–30), cohort mean signed errors converged near zero (+0.81 bpm for OURA; +0.08 bpm for WHOOP), yet individual MAEs remained persistently elevated at 8.37 and 7.56 bpm, with 32.6% of OURA and 24.4% of WHOOP readings exceeding ±10 bpm error. Conclusions: The Deep Health® plantar platform achieved sub-millimeter-of-mercury agreement with manual auscultation during quiet stance within the normotensive-to-prehypertensive range studied. Conversely, consumer upper-limb wearables exhibited severe initial algorithmic lag during rapid exertion and substantial participant-level error dispersion during sustained running that was masked by near-zero group mean bias. Furthermore, neither ring nor wrist PPG platforms possess the physical counter-bearing or hydrostatic transduction architecture required for instantaneous, uncalibrated blood pressure measurement.

Authors

Institutions

Publication Details

Journal
Zenodo (CERN European Organization for Nuclear Research)
Published
2026-11-12
DOI
https://doi.org/10.5281/zenodo.24176146
Primary Topic
Non-Invasive Vital Sign Monitoring
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Hemodynamic Fidelity Across Anatomical Acquisition Sites: Stationary Blood Pressure Concordance and Dynamic Treadmill Heart Rate Tracking of a Plantar Multi-Sensor Platform Versus Finger and Wrist Photoplethysmography in 483 Adults

Kai Weng, Ethan Wei
Zenodo (CERN European Organization for Nuclear Research)
Non-Invasive Vital Sign Monitoring
article

Hemodynamic Fidelity Across Anatomical Acquisition Sites: Stationary Blood Pressure Concordance and Dynamic Treadmill Heart Rate Tracking of a Plantar Multi-Sensor Platform Versus Finger and Wrist Photoplethysmography in 483 Adults

Kai Weng, Ethan Wei
article en

Abstract

ABSTRACTBackground: Resting agreement between wearable sensors and clinical reference standards does not establish fidelity during dynamic cardiovascular exertion, nor do software-imported blood pressure logs represent direct hemodynamic sensing. This study evaluated stationary blood pressure (BP) and heart rate (HR) agreement of a fixed ground-contact plantar platform alongside resting and dynamic treadmill HR tracking of finger-worn (OURA) and volar wrist-worn (WHOOP) photoplethysmography (PPG) devices in an established user cohort. Methods: In 483 adults (267 women, 216 men) recruited from active personal owners with mature device baselines, the Deep Health® plantar platform (model PH-CS20C3) was compared concurrently against manual brachial auscultation and 60-second radial pulse counts during quiet stance. OURA and WHOOP HR outputs were evaluated at rest and across a 30-minute treadmill protocol (8:00 min/mile [7.5 mph], 2.0° incline) sampled at baseline and every 3 minutes (5,313 paired observations per wearable). Agreement was assessed via mean bias (MD), standard deviation of differences (SDD), Bland–Altman 95% limits of agreement (LoA), mean absolute error (MAE), root mean square error (RMSE), Lin’s concordance correlation coefficient (CCC), and 10,000-iteration participant-level bootstrap confidence intervals. Wearable BP entries were excluded from primary accuracy testing because audit confirmed they reflected static external cuff imports or uncalibrated overnight estimates rather than contemporaneous sensor measurements. Results: Stationary plantar SBP and DBP biases were both −0.01 mmHg, with SDDs of 1.09 mmHg (95% LoA: −2.15 to +2.13 mmHg; CCC = 0.9890) and 0.75 mmHg (95% LoA: −1.48 to +1.45 mmHg; CCC = 0.9932), respectively, with 93.8% of SBP and 99.6% of DBP readings falling within ±2.0 mmHg of manual auscultation. Stationary plantar HR bias was +0.04 bpm (MAE = 0.34 bpm), whereas stationary OURA and WHOOP HR MAEs were 8.50 and 13.85 bpm. During treadmill exercise, reference HR surged by +60.38 bpm at minute 3 (145.51 bpm), whereas OURA and WHOOP lagged at 134.04 bpm (MD = −11.47 bpm) and 135.23 bpm (MD = −10.28 bpm). Across minutes 3–30, overall exercise MAE was 8.43 bpm (95% CI: 8.27–8.59) for OURA and 7.66 bpm (95% CI: 7.50–7.81) for WHOOP. During late exercise (minutes 12–30), cohort mean signed errors converged near zero (+0.81 bpm for OURA; +0.08 bpm for WHOOP), yet individual MAEs remained persistently elevated at 8.37 and 7.56 bpm, with 32.6% of OURA and 24.4% of WHOOP readings exceeding ±10 bpm error. Conclusions: The Deep Health® plantar platform achieved sub-millimeter-of-mercury agreement with manual auscultation during quiet stance within the normotensive-to-prehypertensive range studied. Conversely, consumer upper-limb wearables exhibited severe initial algorithmic lag during rapid exertion and substantial participant-level error dispersion during sustained running that was masked by near-zero group mean bias. Furthermore, neither ring nor wrist PPG platforms possess the physical counter-bearing or hydrostatic transduction architecture required for instantaneous, uncalibrated blood pressure measurement.

Zenodo (CERN European Organization for Nuclear Research)
James Cook University Singapore (SG)
Good health and well-being
Non-Invasive Vital Sign Monitoring
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon