Thomas Lilienkamp
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Thomas Lilienkamp has a total of 17 co-authors. Here are the top co-authors by number of shared publications:
Ulrich Parlitz
Max Planck Institute for Dynamics and Self-Organization, Gottingen, Germany
6 shared publications- Ulrich Parlitz
Max Planck Institute for Dynamics and Self-Organization
6 shared publications - Stefan Luther
Max Planck Institute for Dynamics and Self-Organization, Biomedical Physics Group, Goettingen, Germany; University Medical Center Goettingen, Institute of Pharmacology and Toxicology, Goettingen, Germany; German Center for Cardiovascular Research partner site Lower Saxony, Goettingen, Germany; Institute for the Dynamics of Complex Systems, University of Goettingen, Goettingen, Germany
4 shared publications Stefan Luther
Max Planck Institute for Dynamics and Self-Organization, Gottingen, Germany
2 shared publications- Alexander Schlemmer
Max Planck Institute for Dynamics and Self-Organization, 37077 Göttingen, Germany
1 shared publication - Alexandre R. Nieto
Universidad Rey Juan Carlos
1 shared publication
Latest Publications
- Open Access
- 1 September 2026
- Published by AIP Publishing in Chaos: An Interdisciplinary Journal of Nonlinear Science
- Vol. 36 (9)
Deep reinforcement learning to control cardiac arrhythmias via pulse sequences
Life-threatening arrhythmias are the leading cause of sudden cardiac death. The standard clinical treatment involves delivering a high-energy defibrillation shock, which comes along with severe side effects like potential tissue damage or adverse psychological outcomes for patients with implantable cardioverter-defibrillators. We examine whether Reinforcement Learning (RL) can be used to derive low-energy pulse sequences to control the chaotic myocardial excitation dynamics during cardiac arrhythmias. We demonstrate that policies can be trained using numerical simulations and that the resulting multi-pulse protocols offer amplitude reductions of up to 34% compared to state-of-the-art multi-pulse methods. We further show that RL-trained policies increasingly focus on phase singularity dynamics throughout the pulse sequence and that established system observables can be used to explain RL-induced pulse timings. These observations indicate that RL-trained policies are more flexible than previously published protocols and may open up the way toward patient-specific defibrillation strategies.
- 1 September 2026
- Published by AIP Publishing in Chaos: An Interdisciplinary Journal of Nonlinear Science
- Vol. 36 (9)
Open AccessDeep reinforcement learning to control cardiac arrhythmias via pulse sequences
Life-threatening arrhythmias are the leading cause of sudden cardiac death. The standard clinical treatment involves delivering a high-energy defibrillation shock, which comes along with severe side effects like potential tissue damage or adverse psychological outcomes for patients with implantable cardioverter-defibrillators. We examine whether Reinforcement Learning (RL) can be used to derive low-energy pulse sequences to control the chaotic myocardial excitation dynamics during cardiac arrhythmias. We demonstrate that policies can be trained using numerical simulations and that the resulting multi-pulse protocols offer amplitude reductions of up to 34% compared to state-of-the-art multi-pulse methods. We further show that RL-trained policies increasingly focus on phase singularity dynamics throughout the pulse sequence and that established system observables can be used to explain RL-induced pulse timings. These observations indicate that RL-trained policies are more flexible than previously published protocols and may open up the way toward patient-specific defibrillation strategies.
- Open Access
- 15 October 2024
- Published by American Physical Society (APS) in Physical Review E
- Vol. 110 (4) , 044207
Influence of conduction heterogeneities on transient spatiotemporal chaos in cardiac excitable media
Life-threatening cardiac arrhythmias such as ventricular fibrillation are often based on chaotic spiral or scroll wave dynamics which can be self-terminating. In this work, we investigate the influence of conduction heterogeneities on the duration of such chaotic transients in generic models of excitable cardiac media. We observe that low and medium densities of heterogeneities extend the average transient lifetime, while at high densities very long transients, potentially persistent chaos, and periodic attractors occur. Published by the American Physical Society 2024
- 15 October 2024
- Published by American Physical Society (APS) in Physical Review E
- Vol. 110 (4) , 044207
Open AccessInfluence of conduction heterogeneities on transient spatiotemporal chaos in cardiac excitable media
Life-threatening cardiac arrhythmias such as ventricular fibrillation are often based on chaotic spiral or scroll wave dynamics which can be self-terminating. In this work, we investigate the influence of conduction heterogeneities on the duration of such chaotic transients in generic models of excitable cardiac media. We observe that low and medium densities of heterogeneities extend the average transient lifetime, while at high densities very long transients, potentially persistent chaos, and periodic attractors occur. Published by the American Physical Society 2024
- Open Access
- 1 December 2022
- Published by AIP Publishing in Chaos: An Interdisciplinary Journal of Nonlinear Science
- Vol. 32 (12) , 121105
Taming cardiac arrhythmias: Terminating spiral wave chaos by adaptive deceleration pacing
Sequences of weak electrical pulses are considered a promising alternative for terminating ventricular and atrial fibrillations while avoiding strong defibrillation shocks with adverse side effects. In this study, using numerical simulations of four different 2D excitable media, we show that pulse trains with increasing temporal intervals between successive pulses (deceleration pacing) provide high success rates at low energies. Furthermore, we propose a simple and robust approach to calculate inter-pulse spacing directly from the frequency spectrum of the dynamics (for instance, computed based on the electrocardiogram), which can be practically used in experiments and clinical applications.
- 1 December 2022
- Published by AIP Publishing in Chaos: An Interdisciplinary Journal of Nonlinear Science
- Vol. 32 (12) , 121105
Open AccessTaming cardiac arrhythmias: Terminating spiral wave chaos by adaptive deceleration pacing
Sequences of weak electrical pulses are considered a promising alternative for terminating ventricular and atrial fibrillations while avoiding strong defibrillation shocks with adverse side effects. In this study, using numerical simulations of four different 2D excitable media, we show that pulse trains with increasing temporal intervals between successive pulses (deceleration pacing) provide high success rates at low energies. Furthermore, we propose a simple and robust approach to calculate inter-pulse spacing directly from the frequency spectrum of the dynamics (for instance, computed based on the electrocardiogram), which can be practically used in experiments and clinical applications.
- Open Access
- 14 July 2022
- Published by Springer Nature in Scientific Reports
- Vol. 12 (1) , 1-10
Non-monotonous dose response function of the termination of spiral wave chaos
The conventional termination technique of life threatening cardiac arrhythmia like ventricular fibrillation is the application of a high-energy electrical defibrillation shock, coming along with severe side-effects. In order to improve the current treatment reducing these side-effects, the application of pulse sequences of lower energy instead of a single high-energy pulse are promising candidates. In this study, we show that in numerical simulations the dose-response function of pulse sequences applied to two-dimensional spiral wave chaos is not necessarily monotonously increasing, but exhibits a non-trivial frequency dependence. This insight into crucial phenomena appearing during termination attempts provides a deeper understanding of the governing termination mechanisms in general, and therefore may open up the path towards an efficient termination of cardiac arrhythmia in the future.
- 14 July 2022
- Published by Springer Nature in Scientific Reports
- Vol. 12 (1) , 1-10
Open AccessNon-monotonous dose response function of the termination of spiral wave chaos
The conventional termination technique of life threatening cardiac arrhythmia like ventricular fibrillation is the application of a high-energy electrical defibrillation shock, coming along with severe side-effects. In order to improve the current treatment reducing these side-effects, the application of pulse sequences of lower energy instead of a single high-energy pulse are promising candidates. In this study, we show that in numerical simulations the dose-response function of pulse sequences applied to two-dimensional spiral wave chaos is not necessarily monotonously increasing, but exhibits a non-trivial frequency dependence. This insight into crucial phenomena appearing during termination attempts provides a deeper understanding of the governing termination mechanisms in general, and therefore may open up the path towards an efficient termination of cardiac arrhythmia in the future.
- Open Access
- 1 June 2022
- Published by AIP Publishing in Chaos: An Interdisciplinary Journal of Nonlinear Science
- Vol. 32 (6) , 063118
Control of escapes in two-degree-of-freedom open Hamiltonian systems
We investigate the possibility of avoiding the escape of chaotic scattering trajectories in two-degree-of-freedom Hamiltonian systems. We develop a continuous control technique based on the introduction of coupling forces between the chaotic trajectories and some periodic orbits of the system. The main results are shown through numerical simulations, which confirm that all trajectories starting near the stable manifold of the chaotic saddle can be controlled. We also show that it is possible to jump between different unstable periodic orbits until reaching a stable periodic orbit belonging to a Kolmogorov–Arnold–Moser island.
- 1 June 2022
- Published by AIP Publishing in Chaos: An Interdisciplinary Journal of Nonlinear Science
- Vol. 32 (6) , 063118
Open AccessControl of escapes in two-degree-of-freedom open Hamiltonian systems
We investigate the possibility of avoiding the escape of chaotic scattering trajectories in two-degree-of-freedom Hamiltonian systems. We develop a continuous control technique based on the introduction of coupling forces between the chaotic trajectories and some periodic orbits of the system. The main results are shown through numerical simulations, which confirm that all trajectories starting near the stable manifold of the chaotic saddle can be controlled. We also show that it is possible to jump between different unstable periodic orbits until reaching a stable periodic orbit belonging to a Kolmogorov–Arnold–Moser island.
- Open Access
- 5 October 2021
- Published by Springer Nature in Scientific Reports
- Vol. 11 (1) , 1-9
Detecting spiral wave tips using deep learning
The chaotic spatio-temporal electrical activity during life-threatening cardiac arrhythmias like ventricular fibrillation is governed by the dynamics of vortex-like spiral or scroll waves. The organizing centers of these waves are called wave tips (2D) or filaments (3D) and they play a key role in understanding and controlling the complex and chaotic electrical dynamics. Therefore, in many experimental and numerical setups it is required to detect the tips of the observed spiral waves. Most of the currently used methods significantly suffer from the influence of noise and are often adjusted to a specific situation (e.g. a specific numerical cardiac cell model). In this study, we use a specific type of deep neural networks (UNet), for detecting spiral wave tips and show that this approach is robust against the influence of intermediate noise levels. Furthermore, we demonstrate that if the UNet is trained with a pool of numerical cell models, spiral wave tips in unknown cell models can also be detected reliably, suggesting that the UNet can in some sense learn the concept of spiral wave tips in a general way, and thus could also be used in experimental situations in the future (ex-vivo, cell-culture or optogenetic experiments).
- 5 October 2021
- Published by Springer Nature in Scientific Reports
- Vol. 11 (1) , 1-9
Open AccessDetecting spiral wave tips using deep learning
The chaotic spatio-temporal electrical activity during life-threatening cardiac arrhythmias like ventricular fibrillation is governed by the dynamics of vortex-like spiral or scroll waves. The organizing centers of these waves are called wave tips (2D) or filaments (3D) and they play a key role in understanding and controlling the complex and chaotic electrical dynamics. Therefore, in many experimental and numerical setups it is required to detect the tips of the observed spiral waves. Most of the currently used methods significantly suffer from the influence of noise and are often adjusted to a specific situation (e.g. a specific numerical cardiac cell model). In this study, we use a specific type of deep neural networks (UNet), for detecting spiral wave tips and show that this approach is robust against the influence of intermediate noise levels. Furthermore, we demonstrate that if the UNet is trained with a pool of numerical cell models, spiral wave tips in unknown cell models can also be detected reliably, suggesting that the UNet can in some sense learn the concept of spiral wave tips in a general way, and thus could also be used in experimental situations in the future (ex-vivo, cell-culture or optogenetic experiments).
- Open Access
- 8 July 2021
- Published by IOP Publishing in Journal of Physics: Complexity
- Vol. 2 (3) , 035016
Predicting the duration of chaotic transients in excitable media
The spatiotemporal dynamics of excitable media may exhibit chaotic transients. We investigate this transient chaos in the 2D Fenton–Karma model describing the propagation of electrical excitation waves in cardiac tissue and compute the average duration of chaotic transients in dependence on model parameter values. Furthermore, other characteristics like the dominant frequency, the size of the excitable gap, pseudo ECGs, the number of phase singularities and parameters characterizing the action potential duration restitution curve are determined and it is shown that these quantities can be used to predict the average transient time using polynomial regression.
- 8 July 2021
- Published by IOP Publishing in Journal of Physics: Complexity
- Vol. 2 (3) , 035016
Open AccessPredicting the duration of chaotic transients in excitable media
The spatiotemporal dynamics of excitable media may exhibit chaotic transients. We investigate this transient chaos in the 2D Fenton–Karma model describing the propagation of electrical excitation waves in cardiac tissue and compute the average duration of chaotic transients in dependence on model parameter values. Furthermore, other characteristics like the dominant frequency, the size of the excitable gap, pseudo ECGs, the number of phase singularities and parameters characterizing the action potential duration restitution curve are determined and it is shown that these quantities can be used to predict the average transient time using polynomial regression.
- Open Access
- 21 September 2020
- Published by American Physical Society (APS) in Physical Review E
- Vol. 102 (3) , 032219
Susceptibility of transient chimera states
Chaotic dynamics of a dynamical system is not necessarily persistent. If there is (without any active intervention from outside) a transition towards a (possibly nonchaotic) attractor, this phenomenon is called transient chaos, which can be observed in a variety of systems, e.g., in chemical reactions, population dynamics, neuronal activity, or cardiac dynamics. Also, chimera states, which show coherent and incoherent dynamics in spatially distinct regions of the system, are often chaotic transients. In many practical cases, the control of the chaotic dynamics (either the termination or the preservation of the chaotic dynamics) is desired. Although the self-termination typically occurs quite abruptly and can so far in general not be properly predicted, previous studies showed that in many systems a ‘terminal transient phase” (TTP) prior to the self-termination existed, where the system was less susceptible against small but finite perturbations in different directions in state space. In this study, we show that, in the specific case of chimera states, these susceptible directions can be related to the structure of the chimera, which we divide into the coherent part, the incoherent part and the boundary in between. That means, in practice, if self-termination is close we can identify the direction of perturbation which is likely to maintain the chaotic dynamics (the chimera state). This finding improves the general understanding of the state space structure during the TTP, and could contribute also to practical applications like future control strategies of epileptic seizures which have been recently related to the collapse of chimera states.
- 21 September 2020
- Published by American Physical Society (APS) in Physical Review E
- Vol. 102 (3) , 032219
Open AccessSusceptibility of transient chimera states
Chaotic dynamics of a dynamical system is not necessarily persistent. If there is (without any active intervention from outside) a transition towards a (possibly nonchaotic) attractor, this phenomenon is called transient chaos, which can be observed in a variety of systems, e.g., in chemical reactions, population dynamics, neuronal activity, or cardiac dynamics. Also, chimera states, which show coherent and incoherent dynamics in spatially distinct regions of the system, are often chaotic transients. In many practical cases, the control of the chaotic dynamics (either the termination or the preservation of the chaotic dynamics) is desired. Although the self-termination typically occurs quite abruptly and can so far in general not be properly predicted, previous studies showed that in many systems a ‘terminal transient phase” (TTP) prior to the self-termination existed, where the system was less susceptible against small but finite perturbations in different directions in state space. In this study, we show that, in the specific case of chimera states, these susceptible directions can be related to the structure of the chimera, which we divide into the coherent part, the incoherent part and the boundary in between. That means, in practice, if self-termination is close we can identify the direction of perturbation which is likely to maintain the chaotic dynamics (the chimera state). This finding improves the general understanding of the state space structure during the TTP, and could contribute also to practical applications like future control strategies of epileptic seizures which have been recently related to the collapse of chimera states.
- Open Access
- 1 May 2020
- Published by AIP Publishing in Chaos: An Interdisciplinary Journal of Nonlinear Science
- Vol. 30 (5) , 051108
Terminating transient chaos in spatially extended systems
In many real-life systems, transient chaotic dynamics plays a major role. For instance, the chaotic spiral or scroll wave dynamics of electrical excitation waves during life-threatening cardiac arrhythmias can terminate by itself. Epileptic seizures have recently been related to the collapse of transient chimera states. Controlling chaotic transients, either by maintaining the chaotic dynamics or by terminating it as quickly as possible, is often desired and sometimes even vital (as in the case of cardiac arrhythmias). We discuss in this study that the difference of the underlying structures in state space between a chaotic attractor (persistent chaos) and a chaotic saddle (transient chaos) may have significant implications for efficient control strategies in real life systems. In particular, we demonstrate that in the latter case, chaotic dynamics in spatially extended systems can be terminated via a relatively low number of (spatially and temporally) localized perturbations. We demonstrate as a proof of principle that control and targeting of high-dimensional systems exhibiting transient chaos can be achieved with exceptionally small interactions with the system. This insight may impact future control strategies in real-life systems like cardiac arrhythmias.
- 1 May 2020
- Published by AIP Publishing in Chaos: An Interdisciplinary Journal of Nonlinear Science
- Vol. 30 (5) , 051108
Open AccessTerminating transient chaos in spatially extended systems
In many real-life systems, transient chaotic dynamics plays a major role. For instance, the chaotic spiral or scroll wave dynamics of electrical excitation waves during life-threatening cardiac arrhythmias can terminate by itself. Epileptic seizures have recently been related to the collapse of transient chimera states. Controlling chaotic transients, either by maintaining the chaotic dynamics or by terminating it as quickly as possible, is often desired and sometimes even vital (as in the case of cardiac arrhythmias). We discuss in this study that the difference of the underlying structures in state space between a chaotic attractor (persistent chaos) and a chaotic saddle (transient chaos) may have significant implications for efficient control strategies in real life systems. In particular, we demonstrate that in the latter case, chaotic dynamics in spatially extended systems can be terminated via a relatively low number of (spatially and temporally) localized perturbations. We demonstrate as a proof of principle that control and targeting of high-dimensional systems exhibiting transient chaos can be achieved with exceptionally small interactions with the system. This insight may impact future control strategies in real-life systems like cardiac arrhythmias.
- Open Access
- 28 August 2019
- Published by Public Library of Science (PLoS) in PLOS ONE
- Vol. 14 (8) , e0221401
Spontaneous termination of chaotic spiral wave dynamics in human cardiac ion channel models
Chaotic spiral or scroll wave dynamics can be found in diverse systems. In cardiac dynamics, spiral or scroll waves of electrical excitation determine the dynamics during life-threatening arrhythmias like ventricular fibrillation. In numerical studies it was found that chaotic episodes of spiral and scroll waves can be transient, thus they terminate spontaneously. We show in this study that this behavior can also be observed using models which describe the ion channel dynamics of human cardiomyocytes (Bueno-Orovio-Cherry-Fenton model and the Ten Tusscher-Noble-Noble-Panfilov model). For both models we find that the average lifetime of the chaotic transients grows exponentially with the system size. With this behavior, we classify the systems into the group of type-II supertransients. We observe a significant difference of the breakup behavior between the models, which results in a distinct dynamics during the final phase just before the termination. The observation of a (temporally) stable single-spiral state affects the prevailing description of the dynamics of type-II supertransients as being “quasi-stationary” and also the feasibility of predicting the spontaneous termination of the spiral wave dynamics. In the long term, the relation between the breakup behavior of spiral waves and properties of chaotic transients like predictability or average transient lifetime may contribute to an improved understanding and classification of cardiac arrhythmias.
- 28 August 2019
- Published by Public Library of Science (PLoS) in PLOS ONE
- Vol. 14 (8) , e0221401
Open AccessSpontaneous termination of chaotic spiral wave dynamics in human cardiac ion channel models
Chaotic spiral or scroll wave dynamics can be found in diverse systems. In cardiac dynamics, spiral or scroll waves of electrical excitation determine the dynamics during life-threatening arrhythmias like ventricular fibrillation. In numerical studies it was found that chaotic episodes of spiral and scroll waves can be transient, thus they terminate spontaneously. We show in this study that this behavior can also be observed using models which describe the ion channel dynamics of human cardiomyocytes (Bueno-Orovio-Cherry-Fenton model and the Ten Tusscher-Noble-Noble-Panfilov model). For both models we find that the average lifetime of the chaotic transients grows exponentially with the system size. With this behavior, we classify the systems into the group of type-II supertransients. We observe a significant difference of the breakup behavior between the models, which results in a distinct dynamics during the final phase just before the termination. The observation of a (temporally) stable single-spiral state affects the prevailing description of the dynamics of type-II supertransients as being “quasi-stationary” and also the feasibility of predicting the spontaneous termination of the spiral wave dynamics. In the long term, the relation between the breakup behavior of spiral waves and properties of chaotic transients like predictability or average transient lifetime may contribute to an improved understanding and classification of cardiac arrhythmias.