Richard Ewald Rosch

Background: Analysis of SPES responses often relies on averaging repeated stimulation trials to improve signal quality. However, this may obscure clinically relevant trial-to-trial variation. We tested whether explicitly modelling cross-trial dependencies improves localisation of the epileptogenic zone, using concordance with the clinical SOZ as a proxy endpoint, and explored whether resection of model-positive channels is associated with postsurgical seizure freedom. Methods: We developed an interleaved Hierarchical Attention Transformer (HAT), a deep learning architecture that models cross-trial and cross-channel dependencies in SPES responses without averaging. We compared it with two baselines averaging responses or trial embeddings. Models were evaluated with patient-held-out, repeated five-fold cross-validation on SPES data from 35 patients. Robustness to reduced trial availability at inference was assessed by restricting test inputs to 1 or 5 trials. Associations with surgical outcome were assessed using AUROC and patient-level tests on the proportion of model-positive channels resected. Results: The HAT had higher SOZ concordance than the trial-averaged baseline (AUROC 0.762 vs 0.721; mean paired difference 0.041; one-sided 95% lower confidence bound 0.009; Holm-adjusted p = 0.0197). Performance changed little when inference was restricted to 1 trial. Outcome analyses did not provide statistical evidence that seizure-free patients had a higher proportion of model-positive channels resected (AUROC 0.634; p = 0.102). Conclusions: Modelling cross-trial dependencies improved concordance with the SOZ compared with trial-averaged approaches, while remaining robust to reduced trial availability at inference. Associations with postsurgical outcome were inconclusive, consistent with limited sample size and training on SOZ labels rather than outcome-aligned labels. Key points: Cross-trial modelling of SPES responses improved concordance with the clinical SOZ compared with trial-averaged baselines. In a 35-patient cohort, SOZ AUROC was higher than a trial-averaged baseline (0.762 vs 0.721). Performance was stable when inference used as few as 1 trial, with the largest drop in the trial-averaged baseline. Resection of model-positive channels did not provide statistical evidence of an association with seizure freedom (AUROC 0.634; one-sided Mann–Whitney p = 0.102).

Objective: There is a growing synergy between the lines of research on cycles in epilepsy and seizure forecasting. It has been conjectured, for instance, that incorporating information about significant seizure cycles into forecasting algorithms can lead to a better‐than‐chance forecasting performance. However, significance and better‐than‐chance are each typically evaluated against only a single null hypothesis, for example, that forecasts are generated by a Poisson process. We here argue that this should be considered only a first step. Our objective is to demonstrate the importance of testing complementary null hypotheses that represent alternative chance models. Methods: To ensure controlled conditions, we use synthetic data generated from simple mathematical models. Samples drawn from gamma distributions are used to generate sequences of random seizure times and random forecasts. We then determine the strength of cycles as a function of the cycle duration and calculate the sensitivity and fraction of time under alarm obtained for the random forecasting algorithm. In both analyses, we apply numerical, surrogate‐based null‐hypothesis testing methods. In the latter case, this includes a straightforward approach to correcting for multiple testing on nonindependent data. Results: Counterintuitively, the random seizure‐time sequences contain multiple prominent cycles, which are judged highly significant by the Rayleigh test. Moreover, randomly forecasting random seizure times results in a sensitivity of 79% at a fraction of time under alarm of only 42%, clearly outperforming a Poisson‐like predictor. In both cases, however, the flexibility and versatility of surrogate‐based null‐hypothesis tests allow us to successfully reveal that all results can be explained by chance models. Significance: Before reaching conclusions on real cycles in epilepsy, the forecastability of seizures, and genuine capacity of forecasting algorithms, it is essential to test and reject several complementary null hypotheses. Many conclusions might not withstand such rigorous tests, allowing the community to focus on those that do.

Brain function and its failures arise from dynamical patterns of neuronal activity shaped by synaptic neurotransmission. Both neurotransmitter receptor expression and neuronal population dynamics show a remarkable regional variability across the human cortex. We leverage this functional specialisation to characterise the relationship between receptor architectonics and electrophysiological signals. Using dynamic causal modelling (DCM), we fitted neural mass models to a normative set of intracranial EEG data. Subsequently, Bayesian model comparison helped to evaluate whether models improved when equipped with constraints on synaptic connectivity, based on regional neurotransmitter receptor densities. The results show that dynamic causal models generated region‐specific intracranial EEG spectra accurately. Incorporating prior information on normative receptor distributions further improved model evidence, indicating that regional variation in receptor density explains variations in synaptic connectivity and ensuing cortical population dynamics. The output is a cortical atlas of neurobiologically informed intracortical synaptic connectivity parameters. These can serve as empirical priors in future, patient‐specific models. In summary, we show that molecular cortical characteristics—that is, receptor densities—enrich and inform generative, biophysically plausible models of coupled neuronal populations. This work helps to explain regional variations in human electrophysiology, provides a methodological foundation to integrate multimodal data, and serves as a normative resource for future DCM studies of electrophysiology.

Cortico‐cortical evoked potentials (CCEPs) are an active electrophysiological technique used during intracranial electroencephalography to evaluate the effective connectivity and influence of therapeutic stimulation between distinct cortical regions and pinpoint epileptogenic zones (EZs) in patients with epilepsy. Various methodologies have been implemented to analyze CCEPs and characterize the epileptogenic networks for EZ localization. Despite its promise, their interpretation remains challenging due to the large volumes of spatially and temporally complex data generated. Early studies focused largely on qualitative descriptors and predefined, semi‐quantitative features such as waveform morphology and peak latencies. However, these methods are limited by the significant heterogeneity in CCEP waveform conformations across patients and cortical regions. The specific technique used for extraction of features, such as the spectral band power and root mean squared values, remains open to empirical refinement, as does choice of appropriate latency windows, with no consensus reached regarding the optimal approach. Graph theoretical metrics such as degree centrality, betweenness centrality, and clustering coefficients can provide a rich representation of epileptogenic network connectivity. However, these metrics are often abstract and difficult to interpret in a clinical setting or to the non‐expert, and their neuroscientific substrates remains poorly understood. The lack of standardization in stimulation protocol and data‐processing pipelines has further contributed to inconsistency in reported findings. Emerging machine learning approaches have been increasingly applied to CCEP data, offering a more data‐driven and potentially generalizable way to identify electrophysiological biomarkers of the epileptogenic effective connectivity. In this article, we discuss qualitative, quantitative, and spectral features; network‐analytical metrics; and more recently, data driven methodologies aimed at improving the interpretability and clinical utility of CCEP data.

Objective: Pediatric status epilepticus (SE) is a medical emergency associated with significant morbidity. Benzodiazepines (BZPs) are the current first‐line treatment, but do not work in more than one third of children presenting with SE. Animal studies have shown that SE can cause changes in synaptic inhibition signaling that can ultimately lead to BZPs becoming ineffective. However, the relevance of these mechanisms in pediatric patients with SE remains unknown.Methods: To test this hypothesis, we combine clinical electroencephalographic (EEG) recordings with dynamic causal modeling (DCM). This approach allows model‐based inference of cortical synaptic coupling parameters based on EEG recorded across distinct oscillatory states.Results: Our DCM revealed that dynamic changes in inhibitory synaptic coupling explain differences in EEG power spectra associated with BZP treatment responsiveness and guide the transition from ictal to interictal state. Furthermore, in silico simulations demonstrate that there are alternative routes to seizure termination even in cortical circuit models unresponsive to BZPs.Significance: Together, our findings confirm that alterations in synaptic inhibition underlie BZP response during pediatric SE. More broadly, this work further demonstrates the utility of computational modeling to validate insights from basic science in clinically accessible recordings in neurological disorders characterized by abnormal brain states.

Many biological systems display circadian and slow multi‐day rhythms, such as hormonal and cardiac cycles. In patients with epilepsy, these cycles also manifest as slow cyclical fluctuations in seizure propensity. However, such fluctuations in symptoms are consequences of the complex interactions between the underlying physiological, pathophysiological, and external causes. Therefore, identifying an accurate model of the underlying system that governs the multi‐day rhythms allows for a more reliable seizure risk forecast and targeted interventions. The primary aim is to develop a personalized strategy for inferring long‐term trajectories of epileptiform activity and, consequently, seizure risk for individual patients undergoing long‐term ECoG sampling via implantable neurostimulation devices. To achieve this goal, the Hankel alternative view of Koopman (HAVOK) analysis is adopted to approximate a linear representation of nonlinear seizure propensity dynamics. The HAVOK framework leverages Koopman theory and delay‐embedding to decompose chaotic dynamics into a linear system of leading delay‐embedded coordinates driven by the low‐energy coordinate (i.e., forcing). The findings reveal the topology of attractors underlying multi‐day seizure cycles, showing that seizures tend to occur in regions of the manifold with strongly nonlinear dynamics. Moreover, it is demonstrated that the identified system driven by forcings with short periods up to a few days accurately predicts patients' slower multi‐day rhythms, which improves seizure risk forecasting.

The presence of interictal epileptiform discharges on EEG may indicate increased epileptic seizure risk. In highly epileptogenic lesions, such as cortical tubers in tuberous sclerosis, these discharges can be recorded with intracranial stereotactic EEG as part of the evaluation for epilepsy surgery. Yet the network mechanisms that underwrite the generation and spread of these discharges remain poorly understood. Here, we investigate the dynamics of interictal epileptiform discharges using a combination of quantitative analysis of invasive EEG recordings and mesoscale neural mass modelling of cortical dynamics. We first characterise spatially organised local dynamics of discharges recorded from 36 separate tubers in eight patients with tuberous sclerosis. We characterise these dynamics with a set of competing explanatory network models using dynamic causal modelling. Bayesian model comparison of plausible network architectures suggests that the recurrent coupling between neuronal populations within, as well as adjacent to, the tuber core explains the travelling wave dynamics observed in these patient recordings. Our results indicate that tuber cores are the spatial sources of interictal discharges that behave like travelling waves with dynamics most probably explained by locally recurrent tuber–perituberal networks. This view integrates competing theories regarding the pathological organisation of epileptic foci and surrounding cortex in patients with tuberous sclerosis by through coupled oscillator dynamics. This recurrent coupling can explain the spread of ictal dynamics and also provide an explanation interictal discharge spread. In the future, we will explore the possible implications of our findings for epilepsy surgery approaches in tuberous sclerosis. image Key points: Interictal epileptiform discharges (IEDs) are abnormal electrical patterns observed in the brains of people with epilepsy and may indicate seizure risk. In tuberous sclerosis, a condition causing epileptic lesions called cortical tubers, IEDs spread from the tuber core to surrounding brain tissue, forming travelling waves. This study used invasive EEG recordings and mathematical models to identify that recurrent connections between the tuber core and its surroundings explain this wave‐like spread. Further in silico simulations demonstrate that this recurrent network architecture supports both interictal discharges and seizure‐like dynamics under different levels of local inhibition

Neuronal networks must balance the need for stable yet flexible dynamics. This is evident during brain development, when synaptic plasticity during critical windows enables adaptability to changing environments whilst ensuring the stability of population dynamics. The emergence of population dynamics that balance stability and flexibility during development is poorly understood. Here, we investigated developmental brain dynamics in larval zebrafish, using in vivo 2-photon imaging to record single-cell activity across major brain regions from 3-8 days post-fertilisation, a highly plastic period in which hunting behaviours are established. Our findings revealed region-specific trajectories in the development of such dynamic regimes: the telencephalon exhibited increased neuronal excitability and long-range correlations, alongside the emergence of scale invariant avalanche dynamics indicative of enhanced flexibility. Conversely, while other regions showed increased state transitions over development, the telencephalon demonstrated a surprising rise in state stability, characterized by slightly longer dwell times and drastically reduced angular velocity in state space. Remarkably, such rotationally stable dynamics persisted up to 5 seconds into the future, indicating the emergence of strong attractors supporting stability over long timescales. Notably, we observed that telencephalon dynamics were maintained near to but not at a phase transition, thus allowing for robust responses while remaining adaptable to novel inputs. Our results highlight regionally-specific trajectories in the relationship between flexibility and stability, illustrating how developing neuronal populations can self-organize to balance these competing demands. Significance Statement: Brain networks must balance the flexibility to adapt to new stimuli with the need for stability. This trade-off is particularly important during periods of high plasticity in brain development. Our study investigates this balance by recording single-cell activity across the entire brain of developing larval zebrafish. We discovered that brain dynamics become increasingly diverse, characterized by both short and long bursts of activity, reflecting increased flexibility. Simultaneously, we observed the emergence of stable dynamics, linked to consistent activity patterns over time. Using a modelling approach, we showed that this stability was driven by the formation of stable attractors that shape the dynamic trajectories. These findings highlight how population mechanisms can shape the dynamic interplay between flexibility and stability in regional networks in the developing brain.

Seizures that continue for beyond five minutes are classified as status epilepticus (SE) and constitute a medical emergency. Benzodiazepines, the current first-line treatment, attempt to terminate SE by increasing the conductance of chloride-permeable type-A GABA receptors (GABA _A Rs). Despite their widespread use, benzodiazepines are ineffective in over a third of cases. Previous research in animal models has demonstrated that changes in intraneuronal chloride homeostasis and GABA _A R physiology may underlie the development of benzodiazepine resistance in SE. However, there remains a need to understand the effect of these changes at a network level to improve translation into the clinical domain. Therefore, informed by data from human EEG recordings of SE and experimental brain slice recordings, we used a large spiking neural network model that incorporates chloride dynamics to investigate and address the phenomenon of benzodiazepine resistance in SE. We found that the GABA _A R reversal potential (E _GABA ) sets SE-like bursting and determines the response to GABA _A R conductance modulation, with benzodiazepines being anti-seizure at low E _GABA and ineffective or pro-seizure at high E _GABA . The SE-like activity and E _GABA depended on a non-linear relationship between the strength of Cl ^- extrusion and GABA _A R conductance, but not on the initial E _GABA of neurons. Independently controlling Cl ^- extrusion in the pyramidal and interneuronal cell populations revealed the critical role of pyramidal cell Cl ^- extrusion in determining the severity of SE activity and the response to simulated benzodiazepine application. Finally, we demonstrate the model’s utility for considering improved therapeutic approaches for terminating SE in the clinic.

Introduction: New onset refractory status epilepticus (NORSE) is a rare and devastating condition characterised by the sudden onset of refractory status epilepticus (RSE) without an identifiable acute or active structural, toxic, or metabolic cause in an individual without a pre-existing diagnosis of epilepsy. Febrile infection-related epilepsy syndrome (FIRES) is considered a subcategory of NORSE and presents following a febrile illness prior to seizure onset. NORSE/FIRES is associated with high morbidity and mortality in children and adults.Methods and results: In this review we first briefly summarise the reported clinical, paraclinical, treatment and outcome data in the literature. We then report on existing knowledge of the underlying pathophysiology in relation toin vitroandin vivopre-clinical seizure and epilepsy models of potential relevance to NORSE/FIRES.Discussion: We highlight how pre-clinical models can enhance our understanding of FIRES/NORSE and propose future directions for research.