Rapid calibration of atrial electrophysiology models using Gaussian process emulators in the ensemble Kalman filter

Q: What is the clinical evidence from this study?

Population: Atrial fibrillation (computational modeling). Intervention: Ensemble Kalman filter with Gaussian process emulators vs. Markov chain Monte Carlo (MCMC). Primary outcome: Parameter calibration accuracy and computational speed.

Mariya Mamajiwala; Cesare Corrado; Christopher W. Lanyon; Steven A. Niederer; Richard D. Wilkinson; Richard H. Clayton

doi:10.1038/s41598-026-39948-9

February 23, 2026Scientific ReportsOpen Access

Rapid calibration of atrial electrophysiology models using Gaussian process emulators in the ensemble Kalman filter

View Full Paper

Ask AI

Bookmark

Share

Key result

Adapted ensemble Kalman filter enables near-real-time atrial model calibration, predicting AF termination in ~94% of samples.

Figures

Why the study?

Patient-specific models to guide clinical decisions in AF require rapid calibration of physics-based models using routine clinical measurements to be of practical value.

Key Points

Rapidly infer patient-specific electrophysiological parameters from routine clinical observations by formulating model calibration as a static non-linear inverse problem.
Replaced high-fidelity computational simulations with Gaussian process emulators trained on an ensemble of M = 202 left atrial simulations generated via openCARP.
Calibrated tissue parameters across 15 universal atrial coordinate locations using a modified ensemble Kalman filter evaluated across 50 synthetic experiment instances.
Benchmarked estimation accuracy across combinations of S1 local activation times, S2 local activation times, and action potential durations against Markov chain Monte Carlo sampling using 30,000 samples.
Gaussian process emulators achieved a coefficient of determination (R2) greater than 95% across all 45 simulator output variables on unseen test data.
Incorporating S2 and action potential duration measurements consistently reduced prediction root-mean-square errors across the ~300,000 nodes of the left atrial mesh compared to S1 measurements alone.
Ensemble Kalman filter computational time scaled at O(K) under parallelization and sustained convergence as observation density increased, whereas Markov chain Monte Carlo sampling performance deteriorated.

Structured PICO

P

Population

Computational models of left atrial electrophysiology (modified Mitchell-Schaeffer model) for atrial fibrillation

I

Intervention

Ensemble Kalman filter (EnKF) adapted with Gaussian process emulators (GPEs)

C

Comparator

Markov chain Monte Carlo (MCMC) sampling

O

Outcome

Accuracy of parameter calibration (RMSE of predicted S1, S2, and APD outputs)surrogate

Limitations

Assumes the posterior distribution to be approximately Gaussian.
Uses a phenomenological model which does not capture complex non-linearities of full biophysical models.
Current pseudo-dynamics do not inherently enforce physiological constraints on parameters.
Requires generating a surrogate model for each new patient, incurring substantial offline computational cost.
Parameters not influential on S1 output were difficult to estimate without additional measurements
Pilot study only included two representative cases
Limitations around parameter identifiability

Implication

The proposed EnKF method with Gaussian process emulators enables near-real-time, patient-specific calibration of atrial electrophysiology models, offering a potential computational tool to guide AF ablation therapy.

Abstract

Atrial fibrillation (AF) is a common cardiac arrhythmia characterised by disordered electrical activity in the atria. The standard treatment is catheter ablation, which is invasive and irreversible. Recent advances in computational electrophysiology offer the potential for patient-specific models that can be used to guide clinical decisions. To be of practical value, we must be able to rapidly calibrate physics-based models using routine clinical measurements. We pose this calibration task as a static inverse problem, where the goal is to infer spatially homogenous tissue-level electrophysiological parameters from the available observations. To make this tractable, we replace the expensive forward model with Gaussian process emulators (GPEs), and propose a novel adaptation of the ensemble Kalman filter (EnKF) for static non-linear inverse problems. The approach yields parameter samples that can be interpreted as coming from the best Gaussian approximation of the posterior distribution. We compare our results with those obtained using Markov chain Monte Carlo (MCMC) sampling and demonstrate the potential of the approach to enable near-real-time patient-specific calibration, a key step towards predicting outcomes of AF treatment within clinical timescales. The approach is readily applicable to a wide range of static inverse problems in science and engineering.

Discussion

No takes yet. Share an insight, caveat, or question.

Cite This Study

Mamajiwala et al. (2026) studied Atrial fibrillation (computational modeling). Ensemble Kalman filter with Gaussian process emulators vs. Markov chain Monte Carlo (MCMC) was evaluated on Parameter calibration accuracy and computational speed. An adaptation of the ensemble Kalman filter using Gaussian process emulators enabled near-real-time calibration of atrial electrophysiology models, correctly predicting atrial fibrillation termination in at least 94% of posterior samples.

synapsesocial.com/papers/699e9143f5123be5ed04eb28 https://doi.org/10.1038/s41598-026-39948-9

View Full Paper

Ask AI

Bookmark

Share

Rapid calibration of atrial electrophysiology models using Gaussian process emulators in the ensemble Kalman filter

Figures

Key Points

Structured PICO

Limitations

Implication

Abstract

Discussion

Cite This Study

Rapid calibration of atrial electrophysiology models using Gaussian process emulators in the ensemble Kalman filter

Figures

Key Points

Structured PICO

Limitations

Implication

Abstract

Discussion

Cite This Study

Keep reading

Keep reading