pith. sign in

arxiv: 2605.10871 · v2 · pith:5C2ALEVVnew · submitted 2026-05-11 · ⚛️ physics.med-ph · cs.AI· cs.LG

Attractor-Vascular Coupling Theory: Formal Grounding and Empirical Validation for AAMI-Standard Cuffless Blood Pressure Estimation from Smartphone Photoplethysmography

Timothy Oladunni , Farouk Ganiyu Adewumi This is my paper

Pith reviewed 2026-05-20 22:37 UTC · model grok-4.3

classification ⚛️ physics.med-ph cs.AIcs.LG
keywords attractor-vascular coupling theoryphotoplethysmographycuffless blood pressureTakens embeddingAAMI standardcardiac stability indexpulse transit timesmartphone PPG
0
0 comments X

The pith

Cardiac attractor geometry from photoplethysmography encodes blood pressure information sufficient for AAMI-standard cuffless estimation.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper introduces Attractor-Vascular Coupling Theory to demonstrate that the geometric structure of cardiac attractors reconstructed from PPG waveforms carries the information needed to estimate blood pressure at levels meeting clinical standards. It grounds the approach in nonlinear dynamical systems by applying Takens delay embedding to extract attractor morphology features. Two theorems, a proposition, and a corollary establish the formal link to vascular coupling parameters and forecast which features will matter most. Validation uses a LightGBM model on pulse transit time and cardiac stability index from 29,684 windows across 46 subjects under leave-one-subject-out cross-validation and single-point calibration, producing systolic and diastolic mean absolute errors of 2.05 mmHg and 1.67 mmHg. The same accuracy holds when restricting inputs to nine smartphone-camera attractor features alone, confirming that everyday devices can deliver AAMI-compliant performance if the theory is correct.

Core claim

Attractor-Vascular Coupling Theory shows that cardiac attractor geometry obtained through Takens embedding of PPG signals encodes the vascular coupling parameters that set blood pressure values. The theory supplies two theorems, one proposition, and one corollary that justify attractor-derived features for estimation and predict their relative importance. A LightGBM model trained on PTT and CSI attractor features under single-point calibration reaches SBP MAE of 2.05 mmHg and DBP MAE of 1.67 mmHg on 29,684 windows from 46 subjects with LOSO-CV, satisfying AAMI/IEEE SP10 while a PPG-only ablation matches the performance within 0.05 mmHg.

What carries the argument

Attractor-Vascular Coupling Theory, which connects cardiac attractor morphology extracted via Takens delay embedding of PPG to the vascular coupling parameters that determine blood pressure.

If this is right

  • Single-point calibration is enough to reach AAMI-standard accuracy in cuffless blood pressure estimation from PPG.
  • Smartphone PPG signals alone deliver performance matching combined ECG-plus-PPG inputs within 0.05 mmHg.
  • The feature importance hierarchy predicted by the theory is confirmed in the trained model.
  • Calibration produces a 91.5 percent reduction in estimation error relative to the uncalibrated case.
  • Between 70 and 76 percent of individual subjects satisfy AAMI criteria under the reported protocol.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same attractor reconstruction could be applied to other wearable optical signals to estimate additional cardiovascular variables.
  • Deployment in consumer smartphone applications would enable continuous blood pressure tracking outside clinical settings.
  • Validation on ambulatory recordings rather than ICU and surgical data would test whether the claimed performance holds during everyday activity.
  • Sensor hardware could be optimized around attractor-derived metrics to improve signal quality for this type of estimation.

Load-bearing premise

The two theorems, proposition, and corollary correctly establish a predictive link between attractor morphology from PPG and the vascular parameters that control blood pressure, rather than only locating features that correlate after calibration.

What would settle it

An independent replication on a new cohort of at least 46 subjects that applies the identical single-point calibration and LOSO-CV protocol and produces mean absolute error above 5 mmHg for systolic or diastolic pressure would show that the attractor features do not suffice for AAMI-standard estimation.

Figures

Figures reproduced from arXiv: 2605.10871 by Farouk Ganiyu Adewumi, Timothy Oladunni.

Figure 2
Figure 2. Figure 2: Per-subject MAE. Red bars (n = 14) are ICU subjects with vasopressor-induced near-constant or extreme BP. Median: 1.87/1.54 mmHg. DATA AVAILABILITY STATEMENT The BIDMC Waveform Database is publicly available at https://physionet.org/content/bidmc/. VitalDB is publicly avail￾able at https://vitaldb.net. Feature extraction code, trained LightGBM models, and the complete experimental pipeline will be released… view at source ↗
Figure 3
Figure 3. Figure 3: MI feature ranking (SBP; DBP overlaid). Attractor morphology ranks [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗
Figure 3
Figure 3. Figure 3: MI feature ranking (SBP; DBP overlaid) on calibrated residuals. CST [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Ablation study. All four configurations satisfy AAMI. PTT + CSI gap: [PITH_FULL_IMAGE:figures/full_fig_p008_4.png] view at source ↗
read the original abstract

This work proposes Attractor-Vascular Coupling Theory (AVCT), a mathematical framework showing that cardiac attractor geometry encodes blood pressure (BP) information sufficient for AAMI-standard estimation, and validates the theory through a calibrated cuffless BP model using photoplethysmography (PPG). AVCT is grounded in Cardiac Stability Theory and operationalized using Takens delay embedding and attractor morphology extraction. Two theorems, one proposition, and one corollary formally justify the use of PPG attractor features for BP estimation and predict the feature-importance hierarchy. A LightGBM model trained on pulse transit time (PTT) and Cardiac Stability Index (CSI) attractor features under single-point calibration was evaluated using strict leave-one-subject-out cross-validation (LOSO-CV) on 46 subjects from BIDMC ICU (n = 9) and VitalDB surgical data (n = 37), comprising 29,684 windows. The model achieved systolic BP (SBP) mean absolute error (MAE) of 2.05 mmHg and diastolic BP (DBP) MAE of 1.67 mmHg, with correlations r = 0.990 and r = 0.991, satisfying the AAMI/IEEE SP10 requirement of MAE below 5 mmHg. Median per-subject MAE was 1.87/1.54 mmHg, and 70%/76% of subjects individually satisfied AAMI criteria. A PPG-only ablation using nine smartphone attractor features matched the ECG+PPG model within 0.05 mmHg, demonstrating that clinical-grade BP tracking is achievable using only a smartphone camera while surpassing prior generalized LOSO-CV results using fewer sensors. All four AVCT predictions were quantitatively confirmed, with 91.5% error reduction from uncalibrated to calibrated estimation (epsilon_cal = 0.915). Unlike post-hoc explainable AI methods, AVCT predicts features satisfying the architectural faithfulness criterion of the Explainable-AI Trustworthiness (EAT) framework and grounding BP estimation in nonlinear dynamical systems theory.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The manuscript proposes Attractor-Vascular Coupling Theory (AVCT), a dynamical-systems framework asserting that cardiac attractor geometry extracted via Takens embedding from PPG encodes blood-pressure information sufficient for AAMI-standard cuffless estimation. Two theorems, one proposition, and one corollary are presented to justify PPG attractor features (PTT and Cardiac Stability Index) and to predict their importance hierarchy. A LightGBM regressor trained on these features with single-point per-subject calibration is evaluated under strict LOSO-CV on 29,684 windows from 46 subjects (BIDMC ICU n=9, VitalDB surgical n=37), yielding SBP MAE 2.05 mmHg and DBP MAE 1.67 mmHg (r=0.990/0.991), satisfying AAMI/IEEE SP10; a PPG-only ablation is reported to support smartphone applicability, and all four AVCT predictions are stated to be quantitatively confirmed.

Significance. If the AVCT predictions and AAMI-level performance were shown to transfer to smartphone-camera PPG, the work would constitute a notable advance toward ubiquitous, hardware-free clinical-grade BP tracking. The formal grounding in nonlinear dynamics, the explicit confirmation of theory-derived feature hierarchy, the use of LOSO-CV on a sizable multi-center cohort, and the PPG-only ablation that nearly matches the ECG+PPG model are genuine strengths that distinguish the contribution from purely data-driven approaches.

major comments (2)
  1. [Abstract, Title, and Datasets/Validation] Abstract, Title, and § on Datasets/Validation: the central claim and title target AAMI-standard cuffless BP estimation from smartphone photoplethysmography, yet all reported results (including the PPG-only ablation) are obtained exclusively on contact medical-grade PPG from BIDMC ICU and VitalDB surgical monitors. These signals differ systematically from smartphone camera PPG in contact vs. non-contact acquisition, illumination variability, motion artifacts, sampling characteristics, and SNR; consequently the reported MAE values, AAMI compliance rates (70%/76% of subjects), and feature-importance hierarchy remain unverified for the intended acquisition modality.
  2. [AVCT Formal Grounding section (theorems, proposition, corollary)] AVCT Formal Grounding section (theorems, proposition, corollary): while the theory is asserted to establish a causal/predictive link between attractor morphology and vascular-coupling parameters that determine BP, the implemented model relies on a fitted per-subject single-point calibration offset together with LightGBM training on derived features. This raises the question whether the quantitative confirmation of the four predictions and the claimed architectural faithfulness to the EAT framework are driven by the theoretical derivation or by post-hoc empirical fitting.
minor comments (2)
  1. [Results section] Results section: report the exact per-dataset subject and window counts and any exclusion criteria applied to the 29,684 windows so that readers can assess the balance between ICU and surgical cohorts.
  2. [Methods] Methods: specify whether Takens embedding parameters (time delay, embedding dimension) were fixed a priori from theory or selected by cross-validation, and whether this choice was subject-independent.

Simulated Author's Rebuttal

2 responses · 1 unresolved

We thank the referee for the constructive comments, which help clarify the scope of our claims. We respond to each major comment below.

read point-by-point responses

  1. Referee: [Abstract, Title, and Datasets/Validation] Abstract, Title, and § on Datasets/Validation: the central claim and title target AAMI-standard cuffless BP estimation from smartphone photoplethysmography, yet all reported results (including the PPG-only ablation) are obtained exclusively on contact medical-grade PPG from BIDMC ICU and VitalDB surgical monitors. These signals differ systematically from smartphone camera PPG in contact vs. non-contact acquisition, illumination variability, motion artifacts, sampling characteristics, and SNR; consequently the reported MAE values, AAMI compliance rates (70%/76% of subjects), and feature-importance hierarchy remain unverified for the intended acquisition modality.

    Authors: We agree that the validation datasets consist of contact medical-grade PPG signals, which differ from smartphone camera PPG in acquisition method, illumination, motion artifacts, and SNR. The title and abstract frame the work as enabling smartphone PPG because the PPG-only ablation shows that the nine attractor features suffice without ECG and are in principle extractable from camera signals. However, we acknowledge that the reported MAE, AAMI compliance, and feature hierarchy have not been directly tested on smartphone data. We will revise the abstract, title, and datasets/validation section to state that empirical results are obtained on clinical contact PPG, with smartphone applicability supported by the ablation and theory, and we will add an explicit note on the need for future smartphone-specific validation. revision: yes

  2. Referee: [AVCT Formal Grounding section (theorems, proposition, corollary)] AVCT Formal Grounding section (theorems, proposition, corollary): while the theory is asserted to establish a causal/predictive link between attractor morphology and vascular-coupling parameters that determine BP, the implemented model relies on a fitted per-subject single-point calibration offset together with LightGBM training on derived features. This raises the question whether the quantitative confirmation of the four predictions and the claimed architectural faithfulness to the EAT framework are driven by the theoretical derivation or by post-hoc empirical fitting.

    Authors: The AVCT theorems, proposition, and corollary are derived from dynamical-systems analysis of cardiac attractors before any modeling or data fitting occurs; they directly predict the use of PTT and CSI features and their importance ordering for BP estimation. The single-point calibration corrects for subject-specific baseline offsets and is not part of the core theoretical claims. LightGBM is applied only to combine the pre-specified theory-derived features. The quantitative confirmation of all four predictions, including the 91.5% error reduction and the observed feature-importance hierarchy, aligns with the a-priori theoretical expectations. We will expand the Formal Grounding section with explicit mappings from each empirical result back to the corresponding theorem or corollary to make this linkage clearer. revision: partial

standing simulated objections not resolved
  • Direct validation on actual smartphone-camera PPG under realistic conditions (variable illumination, motion, sampling rates) is absent from the current datasets and cannot be performed without new experiments.

Circularity Check

0 steps flagged

No Significant Circularity in AVCT Formal Derivation

full rationale

The paper first presents two theorems, one proposition, and one corollary as formal mathematical results grounded in Cardiac Stability Theory and Takens delay embedding; these steps derive the claimed predictive link and feature-importance hierarchy without reference to fitted model outputs or calibration constants. The subsequent LightGBM training, single-point calibration, and LOSO-CV evaluation on the BIDMC/VitalDB windows constitute an independent empirical test that checks whether the predicted hierarchy and AAMI-level performance materialize. Because the formal steps supply content (specific attractor morphology relations) that is not definitionally identical to the post-training MAE numbers or epsilon_cal reduction, the derivation chain does not reduce to its inputs by construction. The reported 91.5% error reduction is an observed outcome after explicit calibration, not a tautological restatement of the theory.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 2 invented entities

Central claim rests on a newly proposed theory and derived index without machine-checked proofs or external benchmarks; single-point calibration functions as a per-subject free parameter, and Takens embedding is invoked as background mathematics.

free parameters (1)
  • single-point calibration offset
    Per-subject calibration point used to adjust model output for individual vascular differences before applying the attractor features.
axioms (1)
  • standard math Takens delay embedding theorem permits reconstruction of system attractor from scalar PPG time series
    Invoked to extract attractor morphology and Cardiac Stability Index from photoplethysmography signals.
invented entities (2)
  • Attractor-Vascular Coupling Theory (AVCT) no independent evidence
    purpose: Mathematical framework asserting that cardiac attractor geometry encodes sufficient BP information
    Newly introduced theory with two theorems, one proposition, and one corollary.
  • Cardiac Stability Index (CSI) no independent evidence
    purpose: Attractor-derived feature claimed to capture vascular coupling for BP prediction
    Introduced as part of AVCT operationalization alongside PTT.

pith-pipeline@v0.9.0 · 5931 in / 1492 out tokens · 48966 ms · 2026-05-20T22:37:08.305581+00:00 · methodology

Review history (2 revisions) →

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

31 extracted references · 31 canonical work pages · 1 internal anchor

  1. [1]

    2024.Global status report on alcohol and health and treatment of substance use disorders

    World Health Organization, “Global report on hypertension: the race against a silent killer,” 2023. [Online]. Available: https: //www.who.int/publications/i/item/9789240081062

    work page arXiv 2023

  2. [2]

    Global disparities of hypertension prevalence and control,

    K. T. Mills, J. D. Bundy, T. N. Kelly, J. E. Reed, P. M. Kearney, K. Reynolds, J. Chen, and J. He, “Global disparities of hypertension prevalence and control,”Circulation, vol. 134, no. 6, pp. 441–450, 2016

    work page 2016

  3. [3]

    Wearable continuous blood pressure monitoring devices based on pulse wave transit time and pulse arrival time: A review,

    Z.-B. Zhou, T.-R. Cui, D. Li, J.-M. Jian, Z. Li, S.-R. Ji, X. Li, J.-D. Xu, H.-F. Liu, Y . Yang, and T.-L. Ren, “Wearable continuous blood pressure monitoring devices based on pulse wave transit time and pulse arrival time: A review,”Materials, vol. 16, no. 6, p. 2133, 2023

    work page 2023

  4. [4]

    Pulse transit time based continuous cuffless blood pressure estimation: A new extension and a comprehensive evaluation,

    X. Ding, B. P. Yan, Y .-T. Zhang, J. Liu, N. Zhao, and H. K. Tsang, “Pulse transit time based continuous cuffless blood pressure estimation: A new extension and a comprehensive evaluation,”Scientific Reports, vol. 7, no. 1, p. 11554, 2017

    work page 2017

  5. [5]

    Towards ubiquitous blood pressure monitoring via pulse transit time: Theory and practice,

    R. Mukkamala, J.-O. Hahn, O. T. Inan, L. K. Mestha, C.-S. Kim, H. Tøreyen, and M. Forouzanfar, “Towards ubiquitous blood pressure monitoring via pulse transit time: Theory and practice,”IEEE Transac- tions on Biomedical Engineering, vol. 62, no. 8, pp. 1879–1901, 2015

    work page 1901

  6. [6]

    A paralleled CNN and transformer network for PPG-based cuff-less blood pressure estimation,

    Z. Tian, A. Liu, G. Zhu, and X. Chen, “A paralleled CNN and transformer network for PPG-based cuff-less blood pressure estimation,” Biomedical Signal Processing and Control, vol. 99, p. 106741, 2025

    work page 2025

  7. [7]

    Benchmarking and enhancing PPG-based cuffless blood pressure estimation methods,

    N. Mathew, Y . Shen, R. Hu, M. Rahimi, and G. Zouridakis, “Benchmarking and enhancing PPG-based cuffless blood pressure estimation methods,” arXiv preprint arXiv:2602.04725, 2026. [Online]. Available: https://arxiv.org/abs/2602.04725

    work page arXiv 2026

  8. [8]

    ANSI/AAMI SP10:2002/(R)2008: Non-invasive sphygmomanometers,

    Association for the Advancement of Medical Instrumentation, “ANSI/AAMI SP10:2002/(R)2008: Non-invasive sphygmomanometers,” AAMI, Arlington, V A, 2008

    work page 2002

  9. [9]

    Beat-to-beat continuous blood pressure estimation using bidirectional long short-term memory network,

    D.-K. Kim, Y .-T. Kim, H. Kim, and D.-J. Kim, “Beat-to-beat continuous blood pressure estimation using bidirectional long short-term memory network,”Sensors, vol. 21, no. 1, p. 96, 2021

    work page 2021

  10. [10]

    Diagnostic features and potential applications of PPG signal in healthcare: A systematic review,

    M. A. Almarshadet al., “Diagnostic features and potential applications of PPG signal in healthcare: A systematic review,”Healthcare, vol. 10, no. 3, p. 547, 2022

    work page 2022

  11. [11]

    Photoplethysmography and its application in clinical physiological measurement,

    J. Allen, B. Rajbhandari, and A. Murray, “Photoplethysmography and its application in clinical physiological measurement,”Physiological Measurement, vol. 42, no. 1, p. 01TR01, 2021

    work page 2021

  12. [12]

    Cardiac Stability Theory: An Axiomatically Grounded Framework for Continuous Cardiac Health Monitoring via Smartphone Photoplethysmography

    T. Oladunni and F. G. Adewumi, “Cardiac stability theory: An axiomatically grounded framework for continuous cardiac health monitoring via smartphone photoplethysmography,” 2025, manuscript under review. [Online]. Available: https://arxiv.org/abs/2604.23876

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  13. [13]

    Cuff-less differential blood pressure estimation using smart phones,

    A. Chandrasekharet al., “Cuff-less differential blood pressure estimation using smart phones,”IEEE Transactions on Biomedical Engineering, vol. 65, no. 11, pp. 2467–2478, 2018

    work page 2018

  14. [14]

    Fractal dynamics in physiology: alterations with disease and aging,

    A. L. Goldberger, L. A. N. Amaral, J. M. Hausdorff, P. C. Ivanov, C.-K. Peng, and H. E. Stanley, “Fractal dynamics in physiology: alterations with disease and aging,”Proceedings of the National Academy of Sciences, vol. 99, pp. 2466–2472, 2002

    work page 2002

  15. [15]

    W. R. Milnor,Hemodynamics, 2nd ed. Baltimore, MD: Williams & Wilkins, 1989

    work page 1989

  16. [16]

    The arterial Windkessel,

    N. Westerhof, J.-W. Lankhaar, and B. E. Westerhof, “The arterial Windkessel,”Med. Biol. Eng. Comput., vol. 47, pp. 131–141, 2009

    work page 2009

  17. [17]

    Detecting strange attractors in turbulence,

    F. Takens, “Detecting strange attractors in turbulence,” inDynamical Systems and Turbulence, ser. Lecture Notes in Mathematics. Springer, 1981, vol. 898, pp. 366–381

    work page 1981

  18. [18]

    A practical method for calculating largest Lyapunov exponents from small data sets,

    M. T. Rosenstein, J. J. Collins, and C. J. De Luca, “A practical method for calculating largest Lyapunov exponents from small data sets,” Physica D: Nonlinear Phenomena, vol. 65, no. 1–2, pp. 117–134, 1993

    work page 1993

  19. [19]

    Kantz and T

    H. Kantz and T. Schreiber,Nonlinear Time Series Analysis, 1st ed. Cambridge University Press, 1997

    work page 1997

  20. [20]

    A review on the nonlinear dynamical system analysis of electrocardiogram signal,

    S. K. Nayak, A. Bit, A. Dey, B. Mohapatra, and K. Pal, “A review on the nonlinear dynamical system analysis of electrocardiogram signal,” Journal of Healthcare Engineering, vol. 2018, p. 6920420, 2018

    work page 2018

  21. [21]

    Embeddings and delays as derived from quantification of recurrence plots,

    J. P. Zbilut and C. L. Webber, “Embeddings and delays as derived from quantification of recurrence plots,”Physics Letters A, vol. 171, no. 3–4, pp. 199–203, 1992

    work page 1992

  22. [22]

    Physiological time-series analysis using approximate entropy and sample entropy,

    J. S. Richman and J. R. Moorman, “Physiological time-series analysis using approximate entropy and sample entropy,”American Journal of Physiology—Heart and Circulatory Physiology, vol. 278, no. 6, pp. H2039–H2049, 2000

    work page 2000

  23. [23]

    A real-time QRS detection algorithm,

    J. Pan and W. J. Tompkins, “A real-time QRS detection algorithm,”IEEE Transactions on Biomedical Engineering, vol. 32, no. 3, pp. 230–236, 1985

    work page 1985

  24. [24]

    LightGBM: A highly efficient gradient boosting decision tree,

    G. Ke, Q. Meng, T. Finley, T. Wang, W. Chen, W. Ma, Q. Ye, and T.-Y . Liu, “LightGBM: A highly efficient gradient boosting decision tree,” inAdvances in Neural Information Processing Systems, vol. 30, 2017. [Online]. Available: https://proceedings.neurips.cc/paper/ 2017/hash/6449f44a102fde848669bdd9eb6b76fa-Abstract.html

    work page 2017

  25. [25]

    A review of machine learning techniques in photoplethysmography for the non-invasive cuff-less measurement of blood pressure,

    C. El-Hajj and P. A. Kyriacou, “A review of machine learning techniques in photoplethysmography for the non-invasive cuff-less measurement of blood pressure,”Biomedical Signal Processing and Control, vol. 58, p. 101870, 2021

    work page 2021

  26. [26]

    Swin transformer-based blood pressure estimation using higher-order spectral features of ECG and PPG signals,

    S. Kumaret al., “Swin transformer-based blood pressure estimation using higher-order spectral features of ECG and PPG signals,”IEEE Transactions on Instrumentation and Measurement, vol. 72, pp. 1–11, 2023

    work page 2023

  27. [27]

    Cuffless blood pressure estimation using calibrated cardiovascular dynamics in the photoplethysmogram,

    H. Samimi and H. R. Dajani, “Cuffless blood pressure estimation using calibrated cardiovascular dynamics in the photoplethysmogram,” Bioengineering, vol. 9, no. 9, p. 446, 2022

    work page 2022

  28. [28]

    End-to-end personalized cuff-less blood pressure monitoring using ECG and PPG signals,

    B. Suhas, R. S. Srinivasa, Y . M. Saidutta, J. Cho, C.-H. Lee, C. Yang, Y . Shen, and H. Jin, “End-to-end personalized cuff-less blood pressure monitoring using ECG and PPG signals,” inProc. IEEE Int. Conf. Acoustics, Speech Signal Process. (ICASSP), 2024, pp. 2101–2105

    work page 2024

  29. [29]

    Explainable deep neural network for multimodal ECG signals: Intermediate versus late fusion,

    T. Oladunni and E. C. Aneni, “Explainable deep neural network for multimodal ECG signals: Intermediate versus late fusion,”IEEE Access, vol. 13, pp. 202 700–202 736, 2025

    work page 2025

  30. [30]

    PhysioBank, PhysioToolkit, and PhysioNet: Components of a new research resource for complex physiologic sig- nals,

    A. L. Goldbergeret al., “PhysioBank, PhysioToolkit, and PhysioNet: Components of a new research resource for complex physiologic sig- nals,” pp. e215–e220, 2000

    work page 2000

  31. [31]

    VitalDB, a high-fidelity multi-parameter vital signs database in surgical patients,

    H.-C. Leeet al., “VitalDB, a high-fidelity multi-parameter vital signs database in surgical patients,”Scientific Data, vol. 9, p. 279, 2022

    work page 2022