Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

AutoVARP -- a framework for automated reproducible inducibility testing in computational models of cardiac electrophysiology

Paolo Seghetti, Matthias Gsell, Anton Prassk, Martin Bishop, Gernot Plank

TL;DR

AutoVARP tackles the critical lack of reproducibility and standardization in in silico cardiac inducibility testing by delivering an automated workflow built on openCARP, carputils, and forCEPSS. It standardizes inputs and encodes the entire induction protocol into shareable plan files, while a four-stage pipeline ($Pre-Pacing$, $S1$, $S2$, $Maintenance$) and checkpointing drive computational efficiency and reproducibility. The framework demonstrates scalability and reproducibility across large subject cohorts, including simple slab geometries and a complex biventricular geometry, with automated post-processing, visualization, and data bundling for replication. By leveraging open-source components and auditable workflows, AutoVARP enables broader access to rigorous in silico inducibility studies and supports cross-lab validation.

Abstract

Simulations of Cardiac Electrophysiology are gaining momentum beyond basic mechanistic studies, as an approach for supporting clinical decision making. The potential for in silico technologies observed from the research community is immense, with studies demonstrating significantly improved therapeutical outcome with little to no additional burden for patients. Two main factors hinder the translation of these technologies from pure research to applications: virtually no reproducibility of results, and lack of standardized procedures. Inspired by a previously published virtual induction study by Arevalo et al. (2016), We address the issues of reproducibility and standardization providing autoVARP, a framework for standardization of virtual arrhythmia inducibility studies, built upon openCARP and the carputils framework. Standardization relies on the previously published forCEPSS framework and is ensured by defining the whole induction study with input files that can be easily shared. Our approach also ensures numerical efficiency by separating the induction study into four stages: (i) pre-pacing with forCEPSS, (ii) S1 pacing tor each steady state, (iii) S2 induction with different extrastimuli, (iv) testing of sustenance of induced reentries. We demonstrate the approach in a large virtual subject cohort to investigate numerical artifacts that may arise when improper setups are provided to perform virtual induction, and additionally showcase autoVARP in a biventricular mesh. AutoVARP addresses effectively the current gap in standardization and reproducibility of results providing a uniform methodology that can be implemented even by non expert users. AutoVARP is highly scalable and adaptable to markedly different anatomies. Although less flexible than in house implementations it provides automated tools to share setups and does not require re-implementation of any process.

AutoVARP -- a framework for automated reproducible inducibility testing in computational models of cardiac electrophysiology

TL;DR

AutoVARP tackles the critical lack of reproducibility and standardization in in silico cardiac inducibility testing by delivering an automated workflow built on openCARP, carputils, and forCEPSS. It standardizes inputs and encodes the entire induction protocol into shareable plan files, while a four-stage pipeline (, , , ) and checkpointing drive computational efficiency and reproducibility. The framework demonstrates scalability and reproducibility across large subject cohorts, including simple slab geometries and a complex biventricular geometry, with automated post-processing, visualization, and data bundling for replication. By leveraging open-source components and auditable workflows, AutoVARP enables broader access to rigorous in silico inducibility studies and supports cross-lab validation.

Abstract

Simulations of Cardiac Electrophysiology are gaining momentum beyond basic mechanistic studies, as an approach for supporting clinical decision making. The potential for in silico technologies observed from the research community is immense, with studies demonstrating significantly improved therapeutical outcome with little to no additional burden for patients. Two main factors hinder the translation of these technologies from pure research to applications: virtually no reproducibility of results, and lack of standardized procedures. Inspired by a previously published virtual induction study by Arevalo et al. (2016), We address the issues of reproducibility and standardization providing autoVARP, a framework for standardization of virtual arrhythmia inducibility studies, built upon openCARP and the carputils framework. Standardization relies on the previously published forCEPSS framework and is ensured by defining the whole induction study with input files that can be easily shared. Our approach also ensures numerical efficiency by separating the induction study into four stages: (i) pre-pacing with forCEPSS, (ii) S1 pacing tor each steady state, (iii) S2 induction with different extrastimuli, (iv) testing of sustenance of induced reentries. We demonstrate the approach in a large virtual subject cohort to investigate numerical artifacts that may arise when improper setups are provided to perform virtual induction, and additionally showcase autoVARP in a biventricular mesh. AutoVARP addresses effectively the current gap in standardization and reproducibility of results providing a uniform methodology that can be implemented even by non expert users. AutoVARP is highly scalable and adaptable to markedly different anatomies. Although less flexible than in house implementations it provides automated tools to share setups and does not require re-implementation of any process.
Paper Structure (29 sections, 5 figures, 1 table)

This paper contains 29 sections, 5 figures, 1 table.

Table of Contents

  1. Introduction
  2. Methods
  3. Definition of Inputs
  4. Anatomical Meshes
  5. Electrophysiological Properties
  6. Pacing Locations
  7. Pacing Protocol
  8. Study definition and execution
  9. Pre-Pacing stage
  10. S1 stage
  11. S2 stage
  12. Maintenance stage
  13. Post processing options
  14. Results
  15. Setting up a new experiment
  16. ...and 14 more sections

Figures (5)

  • Figure 1: Outline of auto-VARP workflow. A cohort of subject-specific anatomical meshes with appropriate labeling is provided, along with definitions of subject-specific pacing locations and electrophysiological properties characterizing normal healthy myocardium, and impaired border zone tissue (left panels). The inducibility test is executed for each subject in a staged workflow comprising a pre-pacing (PP) stage to compute a limit cycle as a reference condition, a S1 stage to stabilize the activation sequence to a prescribed , a S2 stage to perform the inducibility test with variable coupling intervals, and a final maintenance (MT) stage to monitor reentrant activity and measure duration of maintenance (middle panel). With the protocol following an optimized pipeline (center). Simulation data of the maintenance stage are analyzed to generate tabular views on induction results at the cohort level and for individual subjects and pacing locations. All stages are stitched together to generate movies showing induction and maintenance to facilitate lightweight review and documentation.
  • Figure 2: Complete auto-VARP pipeline. Top: subject to be tested for induction, chosen in a cohort of other subjects. The subject specific .json files define electrodes and tissue configurations to be used in the induction study. The settings are defined in the planfile, and the protocols to be tested are defined in the protocols file, which are general for all the virtual subjects in the cohort. Bottom: each protocol from the protocols file is executed on each subject of the cohort. For each protocol, auto-VARP stages are executed sequentially and checkpoint files are saved for each stage in order to restart simulations efficiently. The figure shows snapshots of the membrane potential at each checkpoint instant, color coded for each stage. Each different S2 interval to be tested is independent of other S2 simulations and is restarted from a checkpoint generated using the largest S2 interval provided. At the end of each S2 simulations, a checkpoint file is saved and used to restart a MT simulation to test if the induction was successful. After the execution of auto-VARP, the time stamps on MT checkpoints are used to establish if the induction was sustained or not.
  • Figure 3: Single cell behavior of the ionic models considered in the auto-VARP induction study. a) At the S1 a marked difference in between healthy and border zone tissue is noted, opening a window of vulnerability. By delivering a premature stimulus within this window, the healthy tissue will be excited but not the border zone tissue, causing a unidirectional conduction block. b) Last two obtained with a of 600m s for 100 beats. At the cellular limit cycle $PCL_{n} = APD_n + DI_{n}$ holds.
  • Figure 4: Example of experiment setup and results of the auto-VARP pipeline applied subject 1mmbz.300um.f90 of the isthmus cohort, and results for the whole cohort. a) setup: eight protocols were defined around the isthmus and used to pace the tissue. Tissue properties were initialized using openCARP and assigned according to the configurations section of the planfile. b) execution of auto-VARP for subject 1mmbz.300um.f90. maps computed during the prepace stage are shown, followed by two representative snapshots of S1 and S2 stages. The MT snapshots show the time when the simulation was interrupted, with the associated time stamp used to determine the outcome. c) table summarizing the outcome of auto-VARP for the whole cohort considered. Red squares indicate electrodes which resulted in sustained reentry
  • Figure 5: example of an auto-VARP execution on an anatomical geometry. Top panel: experimental setup used to define the induction protocol. Central panel: results of auto-VARP for the and S1 stage, computed for every electrode of the setup. We highlight the induction protocol for electrode 6 showing four different S2 intervals (330,340,360,370) delivered three times with a decrement of 15m s. Early stimuli (330,340) fail to capture, whereas late stimuli elicit normal conduction in both healthy tissue and border zone. Unidirectional conduction block was achieved for a S2 interval of 360m s and reveals the reentrant circuit, as assessed checking the exit time of the stage.