arxiv: 2604.03197 · v1 · submitted 2026-04-03 · 💻 cs.LG · physics.comp-ph

Recognition: 2 theorem links

· Lean Theorem

Real-Time Surrogate Modeling for Personalized Blood Flow Prediction and Hemodynamic Analysis

Sokratis J. Anagnostopoulos , George Rovas , Vasiliki Bikia , Theodore G. Papaioannou , Athanase D. Protogerou , Nikolaos Stergiopulos

Authors on Pith no claims yet

Pith reviewed 2026-05-13 20:14 UTC · model grok-4.3

classification 💻 cs.LG physics.comp-ph

keywords surrogate modelinghemodynamic predictionneural networksvirtual cohortcardiac outputarterial pressureparameter estimation1D blood flow

0 comments

The pith

A neural surrogate model predicts patient-specific arterial pressure and cardiac output instantaneously from limited inputs.

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

The paper builds a deep neural network that replaces slow 1-D arterial simulations with real-time predictions of blood pressure and cardiac output for individual patients. It first creates a large virtual cohort by sampling parameters according to the multivariate correlations found in a real clinical dataset, which keeps the generated cases physiologically plausible. Training the network on this cohort lets it screen input combinations instantly and reject those that would produce non-physiological results, cutting the cost of building targeted synthetic datasets. The same model supplies a principled way to choose values for unmeasurable terminal resistances and identifies the smallest set of clinical measurements that still allows accurate cardiac-output estimation. When applied to actual patient records, it recovers central aortic systolic pressure and cardiac output directly.

Core claim

A deep neural surrogate trained on a parametrically generated virtual cohort that respects observed multivariate correlations from clinical data can map patient-specific input parameters to arterial pressure waveforms and cardiac output in a single forward pass, while also furnishing a sampling rule for terminal resistance that reduces uncertainty in unobservable parameters and revealing the minimal clinical measurements sufficient to solve the inverse problem for cardiac output.

What carries the argument

Deep neural surrogate model trained on a virtual cohort derived from multivariate clinical correlations, which maps input parameters directly to hemodynamic outputs.

If this is right

Real-time prediction replaces repeated full simulations for screening large parameter spaces or generating hypertensive subgroups.
Principled sampling of terminal resistance reduces the fraction of discarded synthetic cases and lowers uncertainty in unmeasurable parameters.
The identified minimal measurement set shows which clinical variables carry the information needed to invert for cardiac output.
Direct application to real patient data yields estimates of central aortic systolic pressure and cardiac output without additional invasive measurements.

Where Pith is reading between the lines

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

The same surrogate architecture could be retrained on other 1-D or 3-D vascular domains to accelerate personalized simulations beyond the aorta.
Embedding the model in a clinical workflow would allow immediate feedback on proposed parameter sets during patient intake.
If distribution shift appears in new populations, lightweight fine-tuning on a small local cohort could restore accuracy without regenerating the entire virtual dataset.

Load-bearing premise

The multivariate correlations seen in the Asklepios dataset are representative enough of real physiological variation that a network trained on the resulting virtual cohort will generalize without major error to new clinical cases.

What would settle it

Apply the trained surrogate to an independent clinical cohort with measured cardiac output and central pressures; if the predicted values deviate systematically from the measured ones by more than the error tolerance reported in the paper, the generalization claim does not hold.

Figures

Figures reproduced from arXiv: 2604.03197 by Athanase D. Protogerou, George Rovas, Nikolaos Stergiopulos, Sokratis J. Anagnostopoulos, Theodore G. Papaioannou, Vasiliki Bikia.

**Figure 1.** Figure 1: Real-time prediction workflow: The Asklepios dataset is used to draw physiologically correlated values within the parameter space which can be clinically estimated noninvasively. The dataset is augmented using LHS method and 2000 1-D forward simulations are performed using the 1-D cardiovascular code. The parameter space can be adjusted depending on the physiological range depending on the study. The surr… view at source ↗

**Figure 2.** Figure 2: Training performance: Indicative convergence of the forward surrogate mode. (a) Meansquared error loss over 2000 epochs for training and test sets. The convergence minimum indicates no overfitting and good model generalization. (b) Correlation and Bland-Altman plots: agreement between predicted and measured blood pressures for the brachial DBP (left) and SBP (right) network outputs. 7 [PITH_FULL_IMAGE:fi… view at source ↗

**Figure 3.** Figure 3: Pair-wise parameter sensitivity: The sensitivity heatmap of varying pair-wise parameters (while the rest are held fixed), reveals the parameters with the highest influence on the diastolic/systolic blood pressure are the terminal resistance RT , the CO, the compliance C and the HR. For the pulse pressure (P P), the most dominant parameters are the CO, the RT and RC . The geometry, as well as the terminal c… view at source ↗

**Figure 4.** Figure 4: Pair-wise contour maps: The hemodynamic interactions between each parameter pair (while the rest are held fixed) cause distinct 2-D variations in the resulting pressure fields for both SBP (top), and DBP (bottom). The white points represent the in silico patients used for the training. 3.2. Arterial pressure study To better visualize the physiological domains associated with the most sensitive input parame… view at source ↗

**Figure 5.** Figure 5: Pressure iso-surfaces: Plotting the most sensitive hemodynamic parameters reveal distinct regions of low, normal and hypertensive patients for the DBP (left) and SBP (right). The regions outside this range are non-physiological cases. 3.3. Real population matching So far, we have shown that the surrogate outperforms the full solver in predicting generalized fields of virtual patients in near-real time. We… view at source ↗

**Figure 6.** Figure 6: Population matching: (a) The distribution of naively sampled terminal resistance (RT ) widens the output pressure distributions, even though the rest of the known parameters are sampled directly from the clinical dataset of Asklepios. (b) Constraining only the range of pressures indicates that the 1-D model is able to closely match the clinical distributions of Asklepios for both DBP and SBP, while the dis… view at source ↗

**Figure 7.** Figure 7: CO identification: (a) Using only the noninvasively obtainable parameters (L, D, C, HR, SBP, DBP) yields high correlation but large limits of agreement, reflecting a strong indication of nonuniqueness for the inverse mapping. (b) Introducing an extra pressure reading at the radial artery (b) or the terminal resistance and compliance (RT , CT ) (c), resolves the ambiguity and recovers CO with near-perfect … view at source ↗

**Figure 8.** Figure 8: R-C vs CO grids: (a) Indicative patient contour show that the terminal resistance has a strong monotonic influence on CO. (b) To isolate the effect of CT , we normalize each RT column with its max CO value to reveal that there are at most 4 solutions of the same CT , which constrains the possible combinations even when only CT is given as input along with X. 3.5. Clinical application: aortic noninvasive he… view at source ↗

**Figure 9.** Figure 9: Cardiac hemodynamics prediction: Using only noninvasive inputs X, the surrogate (a) captures the dominant flow across the clinical dataset, while (b) it provides a very accurate estimate for the aortic pressure. 15 [PITH_FULL_IMAGE:figures/full_fig_p015_9.png] view at source ↗

read the original abstract

Cardiovascular modeling has rapidly advanced over the past few decades due to the rising needs for health tracking and early detection of cardiovascular diseases. While 1-D arterial models offer an attractive compromise between computational efficiency and solution fidelity, their application on large populations or for generating large \emph{in silico} cohorts remains challenging. Certain hemodynamic parameters like the terminal resistance/compliance, are difficult to clinically estimate and often yield non-physiological hemodynamics when sampled naively, resulting in large portions of simulated datasets to be discarded. In this work, we present a systematic framework for training machine learning (ML) models, capable of instantaneous hemodynamic prediction and parameter estimation. We initially start with generating a parametric virtual cohort of patients which is based on the multivariate correlations observed in the large Asklepios clinical dataset, ensuring that physiological parameter distributions are respected. We then train a deep neural surrogate model, able to predict patient-specific arterial pressure and cardiac output (CO), enabling rapid a~priori screening of input parameters. This allows for immediate rejection of non-physiological combinations and drastically reduces the cost of targeted synthetic dataset generation (e.g. hypertensive groups). The model also provides a principled means of sampling the terminal resistance to minimize the uncertainties of unmeasurable parameters. Moreover, by assessing the model's predictive performance we determine the theoretical information which suffices for solving the inverse problem of estimating the CO. Finally, we apply the surrogate on a clinical dataset for the estimation of central aortic hemodynamics i.e. the CO and aortic systolic blood pressure (cSBP).

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.

Desk Editor's Note private letter to a colleague

The paper gives a practical pipeline for building a fast neural surrogate on 1D hemodynamics using clinically-derived parameter correlations, but the generalization claim to real data rests on unshown validation.

read the letter

The core contribution here is a straightforward workflow: pull multivariate correlations from the Asklepios dataset to generate a virtual cohort that respects physiological ranges, train a DNN surrogate on the 1D solver outputs, then use the surrogate for quick forward predictions and to screen out bad parameter sets before running expensive simulations. They also show how the same model can help sample terminal resistance and give a sense of what inputs are needed to back out cardiac output. That integrated loop is new enough in the 1D hemodynamics space and directly tackles the practical pain of throwing away large fractions of simulated cases. The real-time screening and the inverse CO step are the parts that could actually change how people build in-silico cohorts or do quick personalization. The abstract is clear on the pipeline and the motivation, and the choice to anchor the cohort in real clinical statistics is the right move rather than sampling from independent ranges. The soft spot is the missing quantitative evidence. We do not see error bars on the surrogate versus the full 1D solver on held-out cases, no direct comparison of input distributions between the Asklepios-derived cohort and the target clinical set, and no domain-shift checks. If the clinical patients sit outside the correlation structure the model saw in training, both the rejection rule and the CO estimates become unreliable. The stress-test concern about distribution shift is real and not addressed in the provided abstract. This is the kind of work that belongs in a methods-focused journal or conference track on computational physiology. A serious referee would push for the missing validation numbers and a clearer statement of how much the surrogate actually speeds things up versus just running the 1D model with better sampling. I would send it to review rather than desk-reject because the framing is coherent and the clinical application is concrete, even if the current evidence is thin.

Referee Report

3 major / 2 minor

Summary. The manuscript presents a framework for generating a virtual patient cohort from multivariate correlations in the Asklepios dataset and training a deep neural network surrogate model to predict patient-specific arterial pressure and cardiac output in real time. The surrogate enables efficient screening of input parameters to reject non-physiological combinations, provides a method for sampling terminal resistance, identifies sufficient information for inverse CO estimation, and is applied to estimate CO and central systolic blood pressure on a clinical dataset.

Significance. If the surrogate demonstrates accurate generalization from the virtual cohort to real clinical data with quantified error metrics, this work could substantially reduce the computational burden of 1D arterial modeling for large cohorts and enable real-time personalized hemodynamic analysis, with implications for clinical decision support in cardiovascular disease monitoring. The integration of data-driven surrogates with physiological constraints from clinical statistics is a promising direction.

major comments (3)

The central claim that the DNN surrogate trained on the Asklepios-derived virtual cohort generalizes to real clinical data for instantaneous arterial pressure and CO prediction (and subsequent inverse CO estimation) lacks reported quantitative validation metrics such as MAE, R², or error bars against the underlying 1D solver on held-out simulated or clinical data; this is load-bearing for both the forward prediction and inverse-problem claims.
No quantitative assessment of distribution shift (e.g., statistical comparison of input parameter distributions such as age, resistance values, or output hemodynamics) is provided between the Asklepios multivariate correlations used for cohort generation and the target clinical dataset, which directly risks invalidating the generalization and non-physiological sample rejection steps.
The assertion that the model's predictive performance determines 'the theoretical information which suffices for solving the inverse problem of estimating the CO' requires explicit methodology details (e.g., feature ablation results or sensitivity analysis on which inputs enable accurate CO recovery) to support the claim; without this, the information-sufficiency conclusion remains unsubstantiated.

minor comments (2)

The abstract contains the notation 'a~priori' which appears to be a typographical error for 'a priori'.
The manuscript would benefit from clearer reporting of any performance metrics with variability measures (standard deviations or confidence intervals) to strengthen interpretability of results.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for their thoughtful and constructive review. The comments highlight important aspects of validation and methodological clarity that strengthen the manuscript. We address each major comment below and have revised the paper to incorporate quantitative metrics, distribution comparisons, and explicit ablation analyses where these were previously insufficiently detailed.

read point-by-point responses

Referee: The central claim that the DNN surrogate trained on the Asklepios-derived virtual cohort generalizes to real clinical data for instantaneous arterial pressure and CO prediction (and subsequent inverse-problem claims) lacks reported quantitative validation metrics such as MAE, R², or error bars against the underlying 1D solver on held-out simulated or clinical data; this is load-bearing for both the forward prediction and inverse-problem claims.

Authors: We agree that explicit quantitative validation metrics were not presented with sufficient detail in the original submission. In the revised manuscript we have added a new validation subsection reporting MAE, RMSE, and R² values (with standard deviations across five independent training runs) for both pressure waveforms and CO on a held-out portion of the virtual cohort. We also include direct comparison against the 1D solver on the same test cases. For the clinical dataset application we now report mean absolute errors and Pearson correlations for estimated CO and central systolic pressure against available reference values, together with error bars. These additions directly support the generalization and inverse-problem claims. revision: yes
Referee: No quantitative assessment of distribution shift (e.g., statistical comparison of input parameter distributions such as age, resistance values, or output hemodynamics) is provided between the Asklepios multivariate correlations used for cohort generation and the target clinical dataset, which directly risks invalidating the generalization and non-physiological sample rejection steps.

Authors: We acknowledge the absence of a formal distribution-shift analysis. The revised manuscript now includes a dedicated supplementary table and figure that compare key marginal and joint distributions (age, body surface area, terminal resistances, and resulting mean arterial pressure) between the Asklepios-derived virtual cohort and the clinical dataset. We report two-sample Kolmogorov-Smirnov statistics and p-values for each parameter, along with overlaid histograms. While the distributions show substantial overlap, we also discuss the modest differences observed in resistance ranges and their implications for the rejection step. revision: yes
Referee: The assertion that the model's predictive performance determines 'the theoretical information which suffices for solving the inverse problem of estimating the CO' requires explicit methodology details (e.g., feature ablation results or sensitivity analysis on which inputs enable accurate CO recovery) to support the claim; without this, the information-sufficiency conclusion remains unsubstantiated.

Authors: The original text described the information-sufficiency conclusion qualitatively from observed performance differences across input configurations. We have expanded the methods and results sections with an explicit feature-ablation study. The revised manuscript now presents CO estimation accuracy (MAE and R²) when the model is given progressively reduced input sets (full pressure waveform, selected pressure points, heart rate only, etc.). These quantitative results identify the minimal sufficient feature set and are used to justify the information-sufficiency statement. revision: yes

Circularity Check

0 steps flagged

No significant circularity in surrogate training and inverse-problem workflow

full rationale

The paper generates a virtual cohort by sampling from multivariate correlations observed in the Asklepios clinical dataset, trains a DNN surrogate on forward 1-D hemodynamic simulations to predict pressure and CO, uses the surrogate for rapid parameter screening and terminal-resistance sampling, and applies it to clinical data for CO/cSBP estimation. The claim that predictive performance assessment determines sufficient information for the inverse CO problem follows directly from the trained forward map without reducing to a self-definitional tautology, a fitted parameter renamed as prediction, or any self-citation chain. No load-bearing step equates the claimed result to its inputs by construction; the workflow remains externally falsifiable via held-out simulation error and clinical validation.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 0 invented entities

The framework rests on the assumption that the Asklepios dataset statistics are representative and that the 1-D arterial model produces ground-truth hemodynamics for training. No new physical entities are introduced.

free parameters (1)

terminal resistance and compliance values
These are sampled from distributions derived from the clinical dataset and then filtered by the surrogate; they function as fitted parameters for each virtual patient.

axioms (1)

domain assumption Multivariate correlations in the Asklepios dataset capture the full physiological range of arterial parameters
Invoked when generating the virtual cohort to ensure physiological realism.

pith-pipeline@v0.9.0 · 5610 in / 1307 out tokens · 18963 ms · 2026-05-13T20:14:23.994430+00:00 · methodology

discussion (0)

Lean theorems connected to this paper

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

IndisputableMonolith/Cost/FunctionalEquation.lean washburn_uniqueness_aczel unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

We then train a deep neural surrogate model, able to predict patient-specific arterial pressure and cardiac output (CO)... The model also provides a principled means of sampling the terminal resistance...
IndisputableMonolith/Foundation/AlphaCoordinateFixation.lean J_uniquely_calibrated_via_higher_derivative unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

Pair-wise parameter sensitivity... dominant factors shaping the pressure response are RT, CO, C and HR

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

24 extracted references · 24 canonical work pages

[1]

Y. Shi, P. Lawford, D. R. Hose, Review of Zero-D and 1-D models of blood flow in the cardiovascular system, BioMedical Engineering OnLine 10 (2011) 33

work page 2011
[2]

Formaggia, D

L. Formaggia, D. Lamponi, A. Quarteroni, One-dimensional models for blood flow in arteries, Journal of Engineering Mathematics 47 (3-4) (2003) 251–276

work page 2003
[3]

S. J. Sherwin, V. Franke, J. Peir´ o, K. H. Parker, One-dimensional modelling of a vascular network in space-time variables, Journal of Engineering Mathematics 47 (3-

work page
[4]

K. S. Matthys, J. Alastruey, J. Peir´ o, A. W. Khir, P. Segers, P. R. Verdonck, K. H. Parker, S. J. Sherwin, Pulse wave propagation in a model human arterial network: Assessment of 1-D numerical simulations against in vitro measurements, Journal of Biomechanics 40 (15) (2007) 3476–3486

work page 2007
[5]

Reymond, F

P. Reymond, F. Merenda, F. Perren, D. Rufenacht, N. Stergiopulos, Validation of a one-dimensional model of the systemic arterial tree, American Journal of Physiology- Heart and Circulatory Physiology 297 (1) (2009) H208–H222

work page 2009
[6]

Alastruey, K

J. Alastruey, K. H. Parker, J. Peir´ o, S. J. Sherwin, Analysing the pattern of pulse waves in arterial networks: A time-domain study, Journal of Engineering Mathemat- ics 64 (2009) 331–351

work page 2009
[7]

Bikia, S

V. Bikia, S. Pagoulatou, B. Trachet, D. Soulis, A. D. Protogerou, T. G. Papaioannou, N. Stergiopulos, Noninvasive cardiac output and central systolic pressure from cuff- pressure and pulse wave velocity, IEEE journal of biomedical and health informatics 24 (7) (2019) 1968–1981

work page 2019
[8]

Laurent, J

S. Laurent, J. Cockcroft, L. Van Bortel, P. Boutouyrie, C. Giannattasio, D. Hayoz, B. Pannier, C. Vlachopoulos, I. Wilkinson, H. Struijker-Boudier, et al., Expert con- sensus document on arterial stiffness: Methodological issues and clinical applications, European Heart Journal 27 (21) (2006) 2588–2605

work page 2006
[9]

L. M. Van Bortel, S. Laurent, P. Boutouyrie, P. Chowienczyk, J. K. Cruickshank, T. De Backer, J. Filipovsk´ y, S. Huybrechts, F. U. S. Mattace-Raso, A. D. Protogerou, et al., Expert consensus document on the measurement of aortic stiffness in daily practice using carotid-femoral pulse wave velocity, Journal of Hypertension 30 (3) (2012) 445–448

work page 2012
[10]

M. R. Pfaller, J. Pham, A. Verma, L. Pegolotti, N. M. Wilson, D. W. Parker, W. Yang, A. L. Marsden, Automated generation of 0D and 1D reduced-order mod- els of patient-specific blood flow, International Journal for Numerical Methods in Biomedical Engineering 38 (10) (2022) e3639. 17

work page 2022
[11]

Pegolotti, M

L. Pegolotti, M. R. Pfaller, N. L. Rubio, K. Ding, R. Brugarolas Brufau, E. Darve, A. L. Marsden, Learning reduced-order models for cardiovascular simulations with graph neural networks, Computers in Biology and Medicine 168 (2024) 107676

work page 2024
[12]

Bikia, G

V. Bikia, G. Rovas, S. Pagoulatou, N. Stergiopulos, Determination of aortic charac- teristic impedance and total arterial compliance from regional pulse wave velocities using machine learning: an in-silico study, Frontiers in bioengineering and biotech- nology 9 (2021) 649866

work page 2021
[13]

Bikia, P

V. Bikia, P. Segers, G. Rovas, S. Pagoulatou, N. Stergiopulos, On the assessment of arterial compliance from carotid pressure waveform, American Journal of Physiology- Heart and Circulatory Physiology 321 (2) (2021) H424–H434

work page 2021
[14]

Bikia, T

V. Bikia, T. G. Papaioannou, S. Pagoulatou, G. Rovas, E. Oikonomou, G. Siasos, D. Tousoulis, N. Stergiopulos, Noninvasive estimation of aortic hemodynamics and cardiac contractility using machine learning, Scientific reports 10 (1) (2020) 15015

work page 2020
[15]

E. R. Rietzschel, M. L. De Buyzere, S. Bekaert, P. Segers, D. De Bacquer, L. Cooman, P. Van Damme, P. Cassiman, M. Langlois, P. Van Oostveldt, et al., Rationale, design, methods and baseline characteristics of the Asklepios study, European Journal of Cardiovascular Prevention & Rehabilitation 14 (2) (2007) 179–191

work page 2007
[16]

M. D. McKay, R. J. Beckman, W. J. Conover, A comparison of three methods for selecting values of input variables in the analysis of output from a computer code, Technometrics 21 (2) (1979) 239–245

work page 1979
[17]

G. J. Langewouters, Visco-elasticity of the human aorta in vitro in relation to pressure and age, Ph.D. thesis, Krips Repro (1982)

work page 1982
[18]

Segers, E

P. Segers, E. Rietzschel, M. De Buyzere, N. Stergiopulos, N. Westerhof, L. Van Bor- tel, T. Gillebert, P. Verdonck, Three-and four-element windkessel models: assessment of their fitting performance in a large cohort of healthy middle-aged individuals, Pro- ceedings of the Institution of Mechanical Engineers, Part H: Journal of Engineering in Medicine 222 ...

work page 2008
[19]

Z. Lu, R. Mukkamala, Continuous cardiac output monitoring in humans by invasive and noninvasive peripheral blood pressure waveform analysis, Journal of Applied Physiology 101 (2) (2006) 598–608

work page 2006
[20]

Wolak, H

A. Wolak, H. Gransar, L. E. Thomson, J. D. Friedman, R. Hachamovitch, A. Gut- stein, L. J. Shaw, D. Polk, N. D. Wong, R. Saouaf, et al., Aortic size assessment by noncontrast cardiac computed tomography: normal limits by age, gender, and body surface area, JACC: Cardiovascular Imaging 1 (2) (2008) 200–209

work page 2008
[21]

Segers, E

P. Segers, E. R. Rietzschel, M. L. D. Buyzere, S. J. Vermeersch, D. D. Bacquer, L. M. V. Bortel, G. D. Backer, T. C. Gillebert, P. R. Verdonck, Noninvasive (input) impedance, pulse wave velocity, and wave reflection in healthy middle-aged men and women, Hypertension 49 (6) (2007) 1248–1255. 18

work page 2007
[22]

Rietzschel, M

E.-R. Rietzschel, M. L. De Buyzere, S. Bekaert, P. Segers, D. De Bacquer, L. Cooman, P. Van Damme, P. Cassiman, M. Langlois, P. van Oostveldt, P. Ver- donck, G. De Backer, T. C. Gillebert, the Asklepios investigators, Rationale, design, methods and baseline characteristics of the asklepios study, European journal of car- diovascular prevention and rehabil...

work page 2007
[23]

Rietzschel, M

E.-R. Rietzschel, M. L. De Buyzere, S. Bekaert, P. Segers, D. De Bacquer, L. Cooman, P. Van Damme, P. Cassiman, M. Langlois, P. van Oostveldt, et al., Rationale, design, methods and baseline characteristics of the asklepios study, European Journal of Preventive Cardiology 14 (2) (2007) 179–191

work page 2007
[24]

T. G. Papaioannou, D. Soulis, O. Vardoulis, A. Protogerou, P. P. Sfikakis, N. Ster- giopulos, C. Stefanadis, First in vivo application and evaluation of a novel method for non-invasive estimation of cardiac output, Medical engineering & physics 36 (10) (2014) 1352–1357. 19

work page 2014