Loading...
Loading...
Contributors:
Dataset URI:
10.5281/zenodo.23058223If you use this dataset, please cite it along with the original publication of Rodero et al. "Computational models show that functional remodeling, not anatomy, reshapes physiological determinants of cardiac performance in hypertrophic cardiomyopathy", Journal of Molecular and Cellular Cardiology (2026) https://doi.org/10.1016/j.yjmcc.2026.08.002. Closed-loop cardiac cycle simulations in patient-specific four-chamber heart models of hypertrophic cardiomyopathy patients (see related publication and datasets). The record holds seven files per patient, where n = 1 to 5 refers to patients HCM1 to HCM5. HCMn_cycle_light.tar.zst contains the simulations, with one folder per sample (HCMn_cycle/cycle_i, where i is the sample number, starting at 0). HCMn_cycle_states.tar.zst contains the single-cell states that initialise the ionic models. HCMn_X.txt, HCMn_Y.txt, HCMn_output_mask_beat_5.txt, HCMn_xlabels.txt and HCMn_ylabels.txt contain the data of the global sensitivity analysis. The .tar.zst files are tar archives compressed with Zstandard (for example, tar --zstd -xf HCM1_cycle_light.tar.zst). One simulation per sample: a loading phase of 100 ms (200 ms in 4156 of the 4649 HCM5 simulations) followed by five heartbeats, with a cycle length of 1017 ms (HCM1), 984 ms (HCM2), 833 ms (HCM3), 789 ms (HCM4) and 968 ms (HCM5). The analysis in the related publication used the fifth beat. The four-chamber cardiac electromechanics model was run using the Cardiac Arrhythmia Research Package (CARP) and is coupled to a closed-loop CircAdapt circulation, and the haemodynamics are recorded as time series: chamber pressures and volumes (cav.LV.csv, cav.RV.csv, cav.LA.csv, cav.RA.csv), valve fluxes and states (valve.*.csv), arterial and venous segment pressures, volumes and flows (tube.*.csv), and peripheral circulation variables (periphery.SR.csv, periphery.PR.csv). Reduced fibre-level time series are provided for stretch, stretch rate, active tension and transmembrane voltage (Lambda.tdat, LambdaDot.tdat, Tension.tdat, vm.tdat). The large per-node spatial field files (.igb), the displacement field (x.dynpt) and, in almost all samples, the intracellular calcium dumps (.bin) and checkpoint (.roe) files have been removed to keep the archive light, but can be provided upon request. The activation time of each node is given in vm_act_seq.dat. imp_region.dat gives the ionic model region of each node, from 0 to 3 in the order of imp_region in parameters.par (lv, rv, atria and passive). The mechanics solver writes the cavity volumes (lv_endo.vol.dat, rv_endo.vol.dat, la_endo.vol.dat and ra_endo.vol.dat, in mL), the tissue volume (tissue.vol.dat, in mm³) and the pressure applied on each boundary listed in mechanic_nbc of parameters.par (lv_endo, rv_endo, la_endo, ra_endo, pericardium, RPVs and SVC, each in a file ending in .nbc_p.dat, in kPa), with the time in ms in the first column. SAN.trc, fascicles_lv.trc, fascicles_rv.trc and Stimulus_3.trc give the time course of the four stimuli (stimulus[0] to stimulus[3] in parameters.par). The solver logs are electrics.log, mechanics.log and cvsys.log (circulation), and mech_nsolve.dat, mech_lsolve.dat and par_stats.dat record the solver iterations. Input parameters and values are also included in each sample folder (parameters.par and cvs_options.json). Each archive contains every simulation that wrote cav.LV.csv, including simulations that stopped before completing five beats: 1077 (HCM1), 931 (HCM2), 1358 (HCM3), 1893 (HCM4) and 4649 (HCM5) simulations. For HCM1, 306 simulations that ran on a different mesh are not included. x.dynpt.tdat is empty in every sample. In HCM2_cycle/cycle_904, cav.LV.csv and parameters.par were damaged on disk and have been rebuilt: cav.LV.csv from the intact rows and the other output files of the same simulation (the volume of the 18 rows from -18 ms to -1 ms is unknown, nan), and parameters.par from the job script and the parameter file of sample 905. In three HCM4 samples (cycle_1763, cycle_1876 and cycle_1953) and four HCM5 samples (cycle_1323, cycle_1802, cycle_2525 and cycle_2539), part of most output files is overwritten with zero (NUL) bytes, so their time series have gaps. In 342 simulations that stopped during the loading phase (25 in HCM1, 3 in HCM2, 12 in HCM3, 6 in HCM4 and 296 in HCM5), the .tdat, .vol.dat and .nbc_p.dat files are empty, and in 332 of them mech_nsolve.dat, mech_lsolve.dat and par_stats.dat are absent. Six sample folders also hold extra files: plots (HCM2 cycle_30, HCM3 cycle_2, HCM5 cycle_1074 and cycle_6632; HCM3 cycle_2 also holds two videos, a zip file and a copy of cav.LA.csv), a CARP documentation file (HCM3 cycle_300, carp_doc.txt) and the unloaded mesh of the sample in CARP format (HCM5 cycle_6632 and cycle_6633, folder mesh/; cycle_6633 also holds a VTK copy and a ParaView camera file, and cycle_6632 a copy of the unloaded coordinates of cycle_6633). Three HCM3 samples (cycle_864, cycle_995 and cycle_1184) still hold checkpoint (.roe) files, and two of them also intracellular calcium dumps (.bin). HCM5_cycle/cycle_4009_stimdur_1 repeats sample 4009 with a stimulus duration of 1 ms instead of 2 ms; it is not counted among the 4649 simulations and still holds its checkpoint (.roe) and calcium dump (.bin) files. The single-cell steady states that initialise the ionic models of each simulation (im_sv_init in parameters.par) are provided in HCMn_cycle_states.tar.zst, with one .sv file per sample for ToRORd_dynCl (left ventricle), ToRORd_dynCl_rv (right ventricle) and JB_COURTEMANCHE (atria), as HCMn_cycle_states/model/i/i_model_LandHumanStress.sv, where HCMn_cycle_states/ stands for the folder states/ in the im_sv_init paths of parameters.par. To facilitate usage, we also provide the input-output pairs that were used for the global sensitivity analyses. Because some simulations did not work or gave non-physiological outputs, we provide a mask file, where 1 means that a specific parameter combination worked and gave physiological outputs, and 0 that it did not. We also provide the labels used to improve readability. HCMn_X.txt holds the full sampled design, 91,706 rows and 46 columns, where row i (counting from 0) holds the inputs of sample i. HCMn_output_mask_beat_5.txt has one row per sample of the analysis, starting at sample 0: 1197 (HCM1), 960 (HCM2), 1400 (HCM3), 2000 (HCM4) and 7000 (HCM5) rows. HCM1_cycle also holds samples 1197 to 1199, which stopped within the first 4 ms and have no row in the mask. HCMn_Y.txt holds the 48 outputs, computed from the fifth beat, of the samples with mask 1, in increasing sample order: 594 (HCM1), 552 (HCM2), 505 (HCM3), 582 (HCM4) and 551 (HCM5) rows. HCMn_xlabels.txt and HCMn_ylabels.txt give the names of the inputs and outputs in column order, and are the same for the five patients.
Keyword(s):
active tension; cardiac cycle; cardiac digital twin; cardiac electromechanics; CARP.
MeSH Term(s):
Grants and Funding:
No Grants or Funding Data Available
Associated Publication(s):
Supplementary Data:
Related Dataset(s):
No Related Dataset Data Available
The National Library of Medicine (NLM) provides information about biomedical datasets. Inclusion on an NLM Web page does not imply endorsement of, or agreement with, the contents by NLM or the National Institutes of Health. Learn more: NLM Web Policies (nih.gov)