{"id":"4186d50f-06e7-441d-9f2c-947808a3062b","arxiv_id":"1908.07522","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"Simulated coronary flow and wall shear stress are most sensitive to uncertain heart-muscle pressure, less to inlet pressure, and little to wall stiffness or branching-law exponent.","lead":"This study ran about 1,200 coronary artery simulations with flexible walls to see how uncertainty in clinical inputs changes predicted blood flow and wall stress. It finds that uncertainty in the heart muscle pressure around the arteries dominates flow variability, a key factor for plaque risk assessment.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 27% flow/TAWSS variability claim is driven by an assumed 10% dPim/dt input (Sec. 3.2) with no clinical support; the headline scales with this arbitrary input, so the quantitative result is conditional on an unmeasured assumption.","rationale":"The central claim examined is the quantitative sensitivity statement that 10% input variability in dPim/dt yields 27% cv in flow and TAWSS. The most load-bearing concern is that this input distribution is assumed, not clinically derived, and the paper's Sec. 5 admits this. I considered other potential concerns: the wall-thickness contradiction in Sec. 2 (h=0.08 mm vs. the cited 1.0±0.2 mm) is likely a unit/typo issue and does not alter the sensitivity structure; the E_s variance inconsistency (0.24 vs. 0.28 MPa in Secs. 3.4/4.6) is minor; and the method-comparison claim is supported by analytic benchmarks with known answers. The Pim_t assumption is the single point where the headline number would change substantially under a different, equally plausible input. Because Eq. 4 contains dPim/dt as a direct source, the output cv is expected to scale near-linearly with input σ, so the 27% figure is a conditional sensitivity rather than a clinically measured finding. The paper is transparent about this limitation, and the sensitivity ranking and UQ methodology remain valuable. Thus the reader's CONDITIONAL verdict is appropriate, and my read does not change it.","tokens_in":23659,"tokens_out":12674,"duration_ms":254606,"concrete_test":"Calibrate the Pim_t input variance using invasive LV pressure recordings from the same catheterization cohort (or published inter-subject dP/dt_max variability) and rerun the 200-sample QMC propagation from Sec. 3.6 with the empirical σ. If the resulting flow/TAWSS cv differs from 27% by more than the ratio of the empirical σ to 10% (e.g., near-linear scaling to ~13.5% for σ=5% or ~54% for σ=20%), the headline is not robust to the input assumption. If no clinical data are available, run the same propagation with σ=5% and σ=20% of max dPim/dt to demonstrate whether the output cv scales linearly with the assumed input; a linear scaling would confirm that the 27% figure is a direct artifact of the unmeasured 10% choice.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's leading quantitative result — that 10% input variability in dPim/dt produces 27% cv in flow and TAWSS (Secs. 4.4 and 5) — rests on an input distribution that is neither measured nor clinically calibrated. Section 3.2 states the process variation is 'σ equal to 10% of the maximum value assumed by Pim,t(t) during the heart cycle,' applied only in systole via a KL expansion. This is not a coefficient of variation in the usual sense, since the mean of a time derivative over a cardiac cycle is near zero; it is a peak-relative standard deviation chosen by the authors. No catheterization data, LV pressure recordings, or literature estimate support this 10% value. Section 5 concedes that 'the distribution of the random inputs were assumed in this study rather than inferred from available clinical data.' Because dPim/dt enters the LPN distal-pressure ODE (Eq. 4) as a direct source term, the output variability in Q and TAWSS is approximately proportional to the input amplitude. Changing the assumed σ from 10% to a clinically realistic value would scale the 27% headline correspondingly. The qualitative ranking — Pim_t is the dominant source of flow/TAWSS variability — may survive, but the numerical magnitude, and the title's promise of 'clinically-derived parametric data uncertainty,' are conditional on an unmeasured parameter. This is not an internal inconsistency; it is a limitation that directly controls the paper's central quantitative claim.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript performs forward uncertainty quantification for a patient-specific left coronary artery model with deformable walls, using an ALE fluid-structure interaction framework and lumped-parameter network outlet boundary conditions. Stochastic inputs are the inlet coronary pressure waveform (modeled from catheterization data via a Karhunen-Loève expansion), the intramyocardial pressure time derivative (assumed 10% peak-relative standard deviation), the morphometry exponent (uniform on 2.4 to 2.8), and the wall Young's modulus (Gaussian with literature-based moments). Uncertainty propagation is carried out with Monte Carlo, quasi-Monte Carlo, stochastic collocation, and multiwavelet stochastic expansion, first on analytic benchmarks and then on the coronary model using 1203 simulations. The main findings are that 7% input coefficient of variation in inlet pressure propagates to about 7% cv in outlet pressure and wall deformation and about 5% cv in flow and wall shear stress; 10% peak-relative variability in the intramyocardial pressure derivative produces up to 27% cv in flow rate and time-averaged wall shear stress; the morphometry exponent has negligible effect; and Young's modulus uncertainty affects wall deformation only, with about 17% cv. The authors conclude that the multiwavelet method is superior to quasi-Monte Carlo and stochastic collocation for this class of problems.","tokens_in":23973,"tokens_out":5810,"duration_ms":55932,"significance":"If the input distributions are accepted, the paper is a useful demonstration of uncertainty quantification in deformable-wall coronary simulations and provides a fair, sample-counted comparison of propagation methods. The strengths include the use of actual intra-coronary catheterization data for the inlet pressure, a boundary-layer mesh convergence study, explicit reporting of the 1203 simulation count, and systematic benchmarks against analytic and nonlinear test problems. The qualitative ranking, namely that intramyocardial pressure uncertainty dominates flow and wall shear stress variability while wall stiffness uncertainty affects mainly wall mechanics, is clinically plausible and worth reporting. However, the leading quantitative claim of 27% cv in flow and TAWSS is conditional on an assumed 10% peak-relative standard deviation for the intramyocardial pressure derivative, and the paper's own limitation statement concedes that the random input distributions were assumed rather than inferred from clinical data. The title's promise of 'clinically-derived parametric data uncertainty' is therefore not supported for the central quantitative result.","major_comments":[{"comment":"The 10% peak-relative standard deviation assigned to the intramyocardial pressure derivative in Sec. 3.2 is a modeling assumption, not an estimate from clinical data. Section 5 explicitly states that 'the distribution of the random inputs were assumed in this study rather than inferred from available clinical data.' Because dPim/dt appears as a direct source term in the distal-pressure ODE, Eq. (4), the reported 27% cv in flow rate and time-averaged wall shear stress (Sec. 4.4) is approximately proportional to this assumed input amplitude. The authors should either estimate this input from clinical measurements, calibrate it against available data, or explicitly present the 27% as conditional and include a sensitivity sweep over the input amplitude. As written, the paper's leading quantitative result and the title's 'clinically-derived' claim are not supported for this input.","section":"Secs. 3.2, 4.4, and 5"},{"comment":"There is an internal inconsistency in the wall thickness specification. The text states a uniform wall thickness h=0.08 mm, then says this is 'consistent with' a coronary wall thickness of 1.0±0.2 mm reported in two echocardiographic studies, and 'larger than' a typical wall thickness equal to 10% of the vessel radius. For a left main diameter of 4 mm, 10% of the radius is 0.2 mm, and the cited 1.0 mm value is 12.5 times larger than 0.08 mm. This inconsistency directly affects the wall deformation quantity of interest and the FSI results. Please correct the value or the citations, and assess how sensitive the wall-mechanics conclusions are to this parameter.","section":"Sec. 2"}],"minor_comments":[{"comment":"The phrase '10% cv in the intramyocardial pressure' is misleading because a coefficient of variation is conventionally defined relative to a nonzero mean, whereas the mean of dPim/dt over a cardiac cycle is near zero; the quantity used is a peak-relative standard deviation and should be labeled as such throughout.","section":"Sec. 3.2"},{"comment":"The Young's modulus input is specified as σ[Es]=0.24 MPa in Sec. 3.4 but as Es∼N(1.48, 0.28^2) in Sec. 4.6; the reported 17% cv in wall deformation is consistent with neither value exactly (16.2% or 18.9%). Please reconcile the two specifications.","section":"Secs. 3.4 and 4.6"},{"comment":"The functions in Eq. (10) are the Sobol' functions, but the text describes the test as a 'ten dimensional sine response surface'; reword to avoid confusion with the sinusoidal benchmark in Eq. (8).","section":"Sec. 4.1"},{"comment":"The name 'Kraichnan-Orzag' is misspelled; it should be 'Kraichnan-Orszag'.","section":"Sec. 4.2"},{"comment":"The sentence 'The MW showed but showed the best performance on discontinuous response surfaces' contains a typographical repetition, and the abstract's broad claim that multiwavelet expansion is 'superior' to stochastic collocation should be qualified because stochastic collocation outperforms all methods on the smooth sine benchmark (Fig. 7).","section":"Sec. 5"},{"comment":"The selection of the correlation length lc=T/2 is stated after 'examining the covariances for the six patients,' but no quantitative comparison of candidate correlation lengths is shown; please provide the covariance fit or the KL eigenvalue decay to justify this choice.","section":"Sec. 3.1"}],"recommendation":"major_revision","confidential_remarks":"To the editor: This is a forward UQ study with a clean propagation framework and no circularity. The main concern is that the headline 27% cv result is driven by an assumed amplitude for the intramyocardial pressure derivative, which the authors themselves acknowledge in Sec. 5; this needs to be reframed as conditional or supported by a sensitivity analysis. The wall-thickness inconsistency in Sec. 2 is an internal error that must be fixed. The methods comparison appears sound, and the paper fits the journal's scope. I do not see grounds for rejection, but the requested revisions are substantive rather than purely editorial."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: this paper is worth engaging with. It does real work that hasn't been done before — UQ on a patient-specific left coronary model with ALE deformable walls, using catheterization-derived pressure uncertainty and comparing MC, QMC, stochastic collocation, and multiwavelet expansion on both analytic benchmarks and the actual model. The mesh convergence study is careful, the 1203 simulations are reported, and the method comparison plots are informative. The claim that multiwavelet converges faster than QMC on this problem is plausible and supported by the plots, even though the method is the authors' own; they also show SC failing on discontinuities, which is consistent with the literature.\n\nThe soft spot is exactly the one the stress-test flagged. The paper's largest quantitative result — 10% variability in dPim/dt producing up to 27% cv in flow and TAWSS — rests on an input distribution that is assumed, not measured. Section 3.2 says the standard deviation is 10% of the maximum absolute value of the baseline dPim/dt, applied only during systole, and Section 5 admits the input distributions were assumed rather than inferred from clinical data. This is not a coefficient of variation in the usual sense because the mean of the derivative over a cycle is near zero. Since dPim/dt enters the distal-pressure ODE as a source term, the output cv scales roughly proportionally to the input amplitude. So the 27% number should be treated as conditional on a plausible but unvalidated choice. The qualitative ranking — Pim_t dominates flow and TAWSS variability — probably survives, but the magnitude could change if real clinical data on intramyocardial pressure variability became available.\n\nThere are two smaller issues. The wall thickness is reported as h=0.08 mm while the text says it is consistent with studies reporting 1.0±0.2 mm; that is off by an order of magnitude and needs correction. And no model/data artifacts are provided, so exact reproduction is not possible from the paper alone.\n\nWho is this for? Anyone doing UQ in cardiovascular simulation, especially with FSI, and people who need sensitivity rankings for coronary boundary conditions. The paper deserves a serious referee: the execution is careful, the literature is handled honestly, and the limitations are acknowledged. I would send it out but ask the authors to add a sensitivity analysis around the assumed Pim_t amplitude, fix the wall thickness inconsistency, and consider making the sub-model and scripts available.","headline":"Solid forward UQ on a deformable coronary model, but the headline 27% flow/TAWSS variability is driven by an assumed 10% intramyocardial pressure input, so the magnitude is conditional even if the ranking is likely right.","tokens_in":24473,"tokens_out":2932,"would_cite":true,"duration_ms":26474,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"A 10 percent uncertainty in the heart muscle's squeezing pressure can spread simulated coronary flow and wall shear stress by 27 percent.","keywords":["coronary circulation","uncertainty quantification","fluid-structure interaction","patient-specific simulation","intramyocardial pressure","Karhunen-Loève expansion","multi-wavelet stochastic expansion","wall shear stress"],"falsifier":"An independent clinical estimate of the systolic time-derivative of intramyocardial pressure variability, propagated through the same left coronary model, would confirm or replace the central 27% figure; if the true coefficient of variation is, for example, 5% instead of 10%, the flow and wall shear stress coefficient of variation would be roughly half the reported value.","tokens_in":23478,"feed_emoji":"🫀","tokens_out":5649,"duration_ms":52704,"temperature":0.7,"pith_summary":"This paper tries to establish how much uncertainty in the input parameters of a patient-specific coronary simulation with deformable walls translates into uncertainty in clinically relevant outputs such as pressure, flow, wall shear stress, and wall deformation. It claims that uncertainty in the inlet pressure waveform passes through almost linearly, while uncertainty in the intramyocardial pressure derivative is the dominant driver of flow and wall shear stress variability, with a 10% input coefficient of variation producing up to 27% output variability. It also claims that wall stiffness uncertainty affects only wall deformation and that the morphometry exponent has little influence on any output. The practical consequence is that deterministic coronary simulations, especially of wall shear stress, should be reported with confidence intervals because one boundary-condition input alone can induce a spread of roughly one quarter in those values.","feed_headline":"10% uncertainty in heart-muscle pressure swings coronary flow 27%","feed_subtitle":"Patient-specific simulations show which input uncertainties matter most for flow, shear stress, and wall motion.","key_machinery":"The mechanism is a sub-modeled left coronary artery with Arbitrary-Lagrangian-Eulerian fluid-structure interaction, coupled at six outlets to lumped-parameter coronary boundary conditions that include the intramyocardial pressure and its time derivative. A pulsatile pressure is prescribed at the inlet, so flow is driven by the pressure difference between the inlet and the downstream intramyocardial pressure. Each uncertain input is represented by random variables: inlet pressure and the intramyocardial pressure derivative through Karhunen-Loève expansion of Gaussian processes, Young's modulus through a Gaussian distribution from tensile-test data, and the morphometry exponent through a uniform distribution. Uncertainty is propagated by Monte Carlo, quasi-Monte Carlo, stochastic collocation, and a multi-wavelet stochastic expansion that adaptively refines the stochastic domain. The load-bearing lever is the 10% systolic perturbation of the intramyocardial pressure derivative, because it directly modifies the pressure gradient that drives coronary flow and therefore produces the reported 27% variability in flow and wall shear stress.","core_discovery":"The paper's central quantitative discovery is a separation of uncertainty transmission paths in a left coronary artery sub-model with deformable walls. A 7% coefficient of variation in the inlet pressure waveform, measured from repeated catheterization data in six patients, transmits almost unchanged, about 7%, to outlet pressure and wall deformation, and about 5% to flow rate and time-averaged wall shear stress. A 10% coefficient of variation assumed for the systolic time derivative of intramyocardial pressure produces up to 27% coefficient of variation in flow rate and wall shear stress, while leaving pressure and deformation variability below 3%. Young's modulus uncertainty, modeled as a Gaussian with 17% coefficient of variation from human coronary tensile-test data, affects only wall deformation, also at about 17%, leaving hemodynamics nearly unchanged. Morphometry exponent uncertainty in the range 2.4 to 2.8 has negligible effects. These results indicate which uncertain inputs must be measured more tightly before coronary simulations can report flow and wall shear stress with clinical confidence.","pith_inferences":["If the reported near-linear transmission of inlet pressure variability holds in stenosed vessels, outlet pressure variability could be approximated directly from inlet measurement variability without a full fluid-structure simulation, a shortcut that could be tested on synthetic stenotic geometries.","Because the intramyocardial pressure derivative was perturbed only during systole, the systolic-window definition is a hidden sensitivity; perturbing the derivative over the full cardiac cycle would test how much of the 27% figure depends on that modeling choice.","The near-decoupling of hemodynamics from wall mechanics suggests a rigid-wall model with identical pressure boundary conditions may reproduce the flow and wall shear stress variability at lower computational cost, which a direct rigid-wall comparison could verify.","The 27% spread in time-averaged wall shear stress implies that plaque-progression risk categories based on a single deterministic value may misclassify patients near thresholds, so reporting the full output distribution would be more clinically informative."],"forward_implications":["If these results transfer to other coronary anatomies, deterministic coronary simulations should be interpreted with confidence intervals: a 10% uncertainty in the systolic intramyocardial pressure derivative alone can spread time-averaged wall shear stress by up to 27% coefficient of variation.","Flow and time-averaged wall shear stress variability track each other, while pressure and wall deformation variability track each other, so one member of each pair can serve as a practical proxy for the other in uncertainty reporting.","Uncertainty in vessel wall stiffness has little bearing on hemodynamic outputs in this small-deformation coronary model, but it must be controlled when wall deformation itself is the quantity of interest.","The multi-wavelet stochastic expansion estimates means and standard deviations accurately with roughly 50 or fewer model evaluations, making uncertainty quantification tractable for deformable coronary sub-models.","Improving the measurement of intramyocardial pressure, rather than inlet pressure, is the bottleneck for accurate flow and wall shear stress predictions.","",""],"supporting_citations":[{"why":"Supplies the human coronary Young's modulus mean and standard deviation used for the Gaussian material-property model.","marker":"[45]"},{"why":"Provides the coronary lumped-parameter network boundary condition structure that includes intramyocardial pressure and its time derivative.","marker":"[46]"},{"why":"Supplies the patient-specific multiscale coronary modeling framework and the morphometry-based resistance distribution approach.","marker":"[70]"},{"why":"Provides the baseline intramyocardial pressure time-derivative waveform and the automated tuning approach used to build the coronary boundary conditions.","marker":"[88]"},{"why":"Prior uncertainty quantification study on coronary bypass grafts with deformable walls that this work extends to an ALE formulation.","marker":"[89]"},{"why":"Supplies the generalized multi-resolution stochastic expansion framework and its implementation parameters used for propagation.","marker":"[74]"},{"why":"Introduces the sparse multiresolution regression approach underlying the multi-wavelet stochastic expansion.","marker":"[76]"},{"why":"Documents variability in arterial stiffness measurements, supporting the stochastic treatment of vessel wall elasticity.","marker":"[33]"}],"fun_headline_variants":["Heart-muscle pressure uncertainty swings coronary flow 27%","Which uncertain inputs move coronary simulations most?","Multiwavelet expansion beats quasi-Monte Carlo for coronary UQ","Patient-specific coronary models: pressure inputs dominate outcomes","Wall stiffness uncertainty only deforms coronary model, not flow"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is the assumed 10% coefficient of variation for the intramyocardial pressure time derivative during systole; this distribution was assumed rather than inferred from clinical data, and it drives the largest reported output variability.","fun_headline_variants_meta":{"raw":{"variants":["Heart-muscle pressure uncertainty swings coronary flow 27%","Which uncertain inputs move coronary simulations most?","Multiwavelet expansion beats quasi-Monte Carlo for coronary UQ","Patient-specific coronary models: pressure inputs dominate outcomes","Wall stiffness uncertainty only deforms coronary model, not flow"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001233,"raw_usage":{"total_tokens":5111,"prompt_tokens":1035,"completion_tokens":4076,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":651,"completion_tokens_details":{"reasoning_tokens":3996}},"tokens_in":651,"tokens_out":4076,"duration_ms":28980,"temperature":1.0,"reasoning_tokens":3996,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T12:14:25.680775+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"An independent clinical estimate of the systolic time-derivative of intramyocardial pressure variability, propagated through the same left coronary model, would confirm or replace the central 27% figure; if the true coefficient of variation is, for example, 5% instead of 10%, the flow and wall shear stress coefficient of variation would be roughly half the reported value.","supporting_citations":[{"cited_title":"Karimi, M","cited_arxiv_id":null,"evidence_quote":"Supplies the human coronary Young's modulus mean and standard deviation used for the Gaussian material-property model."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the coronary lumped-parameter network boundary condition structure that includes intramyocardial pressure and its time derivative."},{"cited_title":"Sankaran, M","cited_arxiv_id":null,"evidence_quote":"Supplies the patient-specific multiscale coronary modeling framework and the morphometry-based resistance distribution approach."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the baseline intramyocardial pressure time-derivative waveform and the automated tuning approach used to build the coronary boundary conditions."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Prior uncertainty quantification study on coronary bypass grafts with deformable walls that this work extends to an ALE formulation."},{"cited_title":"Schiavazzi, G","cited_arxiv_id":null,"evidence_quote":"Supplies the generalized multi-resolution stochastic expansion framework and its implementation parameters used for propagation."},{"cited_title":"Schiavazzi, A","cited_arxiv_id":null,"evidence_quote":"Introduces the sparse multiresolution regression approach underlying the multi-wavelet stochastic expansion."},{"cited_title":"Gow and C","cited_arxiv_id":null,"evidence_quote":"Documents variability in arterial stiffness measurements, supporting the stochastic treatment of vessel wall elasticity."}],"review_version":1}