The use of computational models in the management and prognosis of refractory epilepsy: A critical evaluation

NeoCoMM: A neocortical neuroinspired computational model for the reconstruction and simulation of epileptiform events

BACKGROUND

Understanding the pathophysiological dynamics that underline Interictal Epileptiform Events (IEEs) such as epileptic spikes, spike-and-waves or High-Frequency Oscillations (HFOs) is of major importance in the context of neocortical refractory epilepsy, as it paves the way for the development of novel therapies. Typically, these events are detected in Local Field Potential (LFP) recordings obtained through depth electrodes during pre-surgical investigations. Although essential, the underlying pathophysiological mechanisms for the generation of these epileptic neuromarkers remain unclear. The aim of this paper is to propose a novel neurophysiologically relevant reconstruction of the neocortical microcircuitry in the context of epilepsy. This reconstruction intends to facilitate the analysis of a comprehensive set of parameters encompassing physiological, morphological, and biophysical aspects that directly impact the generation and recording of different IEEs.

METHOD

a novel microscale computational model of an epileptic neocortical column was introduced. This model incorporates the intricate multilayered structure of the cortex and allows for the simulation of realistic interictal epileptic signals. The proposed model was validated through comparisons with real IEEs recorded using intracranial stereo-electroencephalography (SEEG) signals from both humans and animals. Using the model, the user can recreate epileptiform patterns observed in different species (human, rodent, and mouse) and study the intracellular activity associated with these patterns.

RESULTS

Our model allowed us to unravel the relationship between glutamatergic and GABAergic synaptic transmission of the epileptic neural network and the type of generated IEE. Moreover, sensitivity analyses allowed for the exploration of the pathophysiological parameters responsible for the transitions between these events. Finally, the presented modeling framework also provides an Electrode Tissue Model (ETI) that adds realism to the simulated signals and offers the possibility of studying their sensitivity to the electrode characteristics.

CONCLUSION

The model (NeoCoMM) presented in this work can be of great use in different applications since it offers an in silico framework for sensitivity analysis and hypothesis testing. It can also be used as a starting point for more complex studies.

Machine Learning and Artificial Intelligence Applications to Epilepsy: a Review for the Practicing Epileptologist

PURPOSE OF REVIEW

Machine Learning (ML) and Artificial Intelligence (AI) are data-driven techniques to translate raw data into applicable and interpretable insights that can assist in clinical decision making. Some of these tools have extremely promising initial results, earning both great excitement and creating hype. This non-technical article reviews recent developments in ML/AI in epilepsy to assist the current practicing epileptologist in understanding both the benefits and limitations of integrating ML/AI tools into their clinical practice.

RECENT FINDINGS

ML/AI tools have been developed to assist clinicians in almost every clinical decision including (1) predicting future epilepsy in people at risk, (2) detecting and monitoring for seizures, (3) differentiating epilepsy from mimics, (4) using data to improve neuroanatomic localization and lateralization, and (5) tracking and predicting response to medical and surgical treatments. We also discuss practical, ethical, and equity considerations in the development and application of ML/AI tools including chatbots based on Large Language Models (e.g., ChatGPT). ML/AI tools will change how clinical medicine is practiced, but, with rare exceptions, the transferability to other centers, effectiveness, and safety of these approaches have not yet been established rigorously. In the future, ML/AI will not replace epileptologists, but epileptologists with ML/AI will replace epileptologists without ML/AI.

Screening of Risk Factors for Poor Prognosis in Patients with Refractory Epilepsy Secondary to Encephalomalacia

Objective

The study was aimed at screening the independent prognostic risk factors for refractory epilepsy associated with encephalomalacia (REAE).

Methods

Patients with REAE treated in the First People's Hospital of Linping District from January 2018 to December 2019 were selected. The prognosis was represented by Engel grading. Clinical data of the patients were collected, including age, sex, BMI, lesion sites, number of lesion sites, lesion size, seizure frequency, epilepsy type, and treatment methods. Independent risk factors for poor prognosis were screened by logistic regression analysis. The receiver operating characteristic curve (ROC) was used to evaluate the prognostic efficacy of independent risk factors.

Results

A total of 48 patients were included in this study, including 31 patients (64.58%) in the good prognosis group and 17 patients (35.42%) in the poor prognosis group. The mean age of the poor prognosis group was higher than that of the good prognosis group (P = 0.002). The proportion of patients with multisite lesions in the poor prognosis group was higher than that in the good prognosis group (P = 0.016). The proportion of patients with cerebral malacia lesion diameter ≥ 3 cm in the poor prognosis group was higher than that in the good prognosis group (P = 0.002). The proportion of patients with attack frequency ≥ 2 times/month in the poor prognosis group was higher than in the good prognosis group (P = 0.002). The proportion of patients receiving surgical treatment in the poor prognosis group was lower than that in the good prognosis group (P < 0.001). Age, number of lesion sites, size of encephalomalacia, and seizure frequency were independent risk factors for the prognosis of patients with REAE (OR > 1, P < 0.05). Surgical treatment was an independent protective factor associated with the prognosis of patients with REAE (OR < 1, P < 0.05). The area under the ROC curve of surgical treatment was 0.83 (P = 0.004). The area under the ROC curve of the size of encephalomalacia was 0.72 (P = 0.008). There was a positive correlation between age and size of encephalomalacia and Engel grade (r > 0, P < 0.05). Surgical treatment was negatively correlated with Engel grade (r < 0, P < 0.05). The number of lesion sites and seizure frequency had no significant correlation with Engel (P > 0.05). The proportion of Engel I patients treated with surgery was higher than that treated with drugs (P = 0.001). The ratio of Engel III and IV patients treated with surgery was lower than that treated with medications (P < 0.05).

Conclusion

Age, number of lesion sites, size of encephalomalacia, and seizure frequency are independent risk factors for the prognosis of patients with REAE. Surgical treatment is an independent prognostic factor for patients with REAE. Surgical treatment can significantly improve patient outcomes.

Interictal dynamic network transitions in mesial temporal lobe epilepsy

OBJECTIVE

To reveal the possible routine of brain network dynamic alterations in patients with mesial temporal lobe epilepsy (mTLE) and to establish a predicted model of seizure recurrence during interictal periods.

METHODS

Seventy-nine unilateral mTLE patients with hippocampal sclerosis and 97 healthy controls from two centers were retrospectively enrolled. Dynamic brain configuration analyses were performed with resting-state functional magnetic resonance imaging (MRI) data to quantify the functional stability over time and the dynamic interactions between brain regions. Relationships between seizure frequency and ipsilateral hippocampal module allegiance were evaluated using a machine learning predictive model.

RESULTS

Compared to the healthy controls, patients with mTLE displayed an overall higher dynamic network, switching mainly in the epileptogenic regions (false discovery rate [FDR] corrected p-FDR < .05). Moreover, the dynamic network configuration in mTLE was characterized by decreased recruitment (intra-network communication), and increased integration (inter-network communication) among hippocampal systems and large-scale higher-order brain networks (p-FDR < .05). We further found that the dynamic interactions between the hippocampal system and the default-mode network (DMN) or control networks exhibited an opposite distribution pattern (p-FDR < .05). Strikingly, we showed that there was a robust association between predicted seizure frequency based on the ipsilateral hippocampal-DMN dynamics model and actual seizure frequency (p-perm < .001).

SIGNIFICANCE

These findings suggest that the interictal brain of mTLE is characterized by dynamical shifts toward unstable state. Our study provides novel insights into the brain dynamic network alterations and supports the potential use of DMN dynamic parameters as candidate neuroimaging markers in monitoring the seizure frequency clinically during interictal periods.