MaTiLDA: An Integrated Machine Learning and Topological Data Analysis Platform for Brain Network Dynamics.

Topological data analysis (TDA) combined with machine learning (ML) algorithms is a powerful approach for investigating complex brain interaction patterns in neurological disorders such as epilepsy. However, the use of ML algorithms and TDA for analysis of aberrant brain interactions requires substantial domain knowledge in computing as well as pure mathematics. To lower the threshold for clinical and computational neuroscience researchers to effectively use ML algorithms together with TDA to study neurological disorders, we introduce an integrated web platform called MaTiLDA. MaTiLDA is the first tool that enables users to intuitively use TDA methods together with ML models to characterize interaction patterns derived from neurophysiological signal data such as electroencephalogram (EEG) recorded during routine clinical practice. MaTiLDA features support for TDA methods, such as persistent homology, that enable classification of signal data using ML models to provide insights into complex brain interaction patterns in neurological disorders. We demonstrate the practical use of MaTiLDA by analyzing high-resolution intracranial EEG from refractory epilepsy patients to characterize the distinct phases of seizure propagation to different brain regions. The MaTiLDA platform is available at: https://bmhinformatics.case.edu/nicworkflow/MaTiLDA.

Machine Learning Interpretability Methods to Characterize Brain Network Dynamics in Epilepsy.

The rapid adoption of machine learning (ML) algorithms in a wide range of biomedical applications has highlighted issues of trust and the lack of understanding regarding the results generated by ML algorithms. Recent studies have focused on developing interpretable ML models and establish guidelines for transparency and ethical use, ensuring the responsible integration of machine learning in healthcare. In this study, we demonstrate the effectiveness of ML interpretability methods to provide important insights into the dynamics of brain network interactions in epilepsy, a serious neurological disorder affecting more than 60 million persons worldwide. Using high-resolution intracranial electroencephalogram (EEG) recordings from a cohort of 16 patients, we developed high accuracy ML models to categorize these brain activity recordings into either seizure or non-seizure classes followed by a more complex task of delineating the different stages of seizure progression to different parts of the brain as a multi-class classification task. We applied three distinct types of interpretability methods to the high-accuracy ML models to gain an understanding of the relative contributions of different categories of brain interaction patterns, including multi-focii interactions, which play an important role in distinguishing between different states of the brain. The results of this study demonstrate for the first time that post-hoc interpretability methods enable us to understand why ML algorithms generate a given set of results and how variations in value of input values affect the accuracy of the ML algorithms. In particular, we show in this study that interpretability methods can be used to identify brain regions and interaction patterns that have a significant impact on seizure events. The results of this study highlight the importance of the integrated implementation of ML algorithms together with interpretability methods in aberrant brain network studies and the wider domain of biomedical research.

Social Service Workers' Use of Social Media to Obtain Client Information: Current Practices and Perspectives on a Potential Informatics Platform.

To gain insight into current use of social-media platforms in human services delivery, we systematically surveyed 172 social-service workers from six agencies in a Midwest US city to gather data about social-media usage among social-service providers, potential challenges and benefits of using social media, and whether a social-media-based informatics platform could be valuable. Quantitative analyses showed that approximately half of participants have used social media to collect client-related information; nearly one-quarter indicated "often" or "nearly daily" use. Adjusting for the effects of worker characteristics, social-media use was associated with the type of agency involved and with increased tenure in social services. Adjusted results also showed that participants' comfort with using the potential application was greater in those agencies substantially involved with investigative/legal work. However, trust in the information collected by the potential application was a stronger, independent predictor of comfort using the tool. Qualitative analyses identified numerous challenges and ethical concerns, and positive and negative aspects of a social-media-based informatics platform. If the platform is to be created, work must be done carefully, fully considering ethical issues rightly raised by social service workers, existing agency policies, and professional standards. Future research should investigate ways to negotiate these complex challenges.

Characterizing Brain Network Dynamics using Persistent Homology in Patients with Refractory Epilepsy.

Epilepsy is a common serious neurological disorder that affects more than 65 million persons worldwide and it is characterized by repeated seizures that lead to higher mortality and disabilities with corresponding negative impact on the quality of life of patients. Network science methods that represent brain regions as nodes and the interactions between brain regions as edges have been extensively used in characterizing network changes in neurological disorders. However, the limited ability of graph network models to represent high dimensional brain interactions are being increasingly realized in the computational neuroscience community. In particular, recent advances in algebraic topology research have led to the development of a large number of applications in brain network studies using topological structures. In this paper, we build on a fundamental construct of cliques, which are all-to-all connected nodes with a k-clique in a graph G (V, E), where V is set of nodes and E is set of edges, consisting of k-nodes to characterize the brain network dynamics in epilepsy patients using topological structures. Cliques represent brain regions that are coupled for similar functions or engage in information exchange; therefore, cliques are suitable structures to characterize the dynamics of brain dynamics in neurological disorders. We propose to detect and use clique structures during well-defined clinical events, such as epileptic seizures, to combine non-linear correlation measures in a matrix with identification of geometric structures underlying brain connectivity networks to identify discriminating features that can be used for clinical decision making in epilepsy neurological disorder.

Epilepsy-Connect: An Integrated Knowledgebase for Characterizing Alterations in Consciousness State of Pharmacoresistant Epilepsy Patients.

Alterations in consciousness state are a defining characteristic of focal epileptic seizures. Consequently, understanding the complex changes in neurocognitive networks which underpin seizure-induced alterations in consciousness state is important for advancement in seizure classification. Comprehension of these changes are complicated by a lack of data standardization; however, the use of a common terminological system or ontology in a patient registry minimizes this issue. In this paper, we introduce an integrated knowledgebase called Epilepsy-Connect to improve the understanding of changes in consciousness states during focal seizures of pharmacoresistant epilepsy patients. This registry catalogues over 809 seizures from 70 patients at University Hospital's Epilepsy Center who were undergoing stereotactic electroencephalography (SEEG) monitoring as part of an evaluation for surgical intervention. Although Epilepsy-Connect focuses on consciousness states, it aims to enable users to leverage data from an informatics platform to analyze epilepsy data in a streamlined manner. Epilepsy-Connect is available at https://bmhinformatics.case.edu/Epilepsyconnect/login/.

Association of Peri-ictal Brainstem Posturing With Seizure Severity and Breathing Compromise in Patients With Generalized Convulsive Seizures.

OBJECTIVE: To analyze the association between peri-ictal brainstem posturing semiologies with postictal generalized electroencephalographic suppression (PGES) and breathing dysfunction in generalized convulsive seizures (GCS). METHODS: In this prospective, multicenter analysis of GCS, ictal brainstem semiology was classified as (1) decerebration (bilateral symmetric tonic arm extension), (2) decortication (bilateral symmetric tonic arm flexion only), (3) hemi-decerebration (unilateral tonic arm extension with contralateral flexion) and (4) absence of ictal tonic phase. Postictal posturing was also assessed. Respiration was monitored with thoracoabdominal belts, video, and pulse oximetry. RESULTS: Two hundred ninety-five seizures (180 patients) were analyzed. Ictal decerebration was observed in 122 of 295 (41.4%), decortication in 47 of 295 (15.9%), and hemi-decerebration in 28 of 295 (9.5%) seizures. Tonic phase was absent in 98 of 295 (33.2%) seizures. Postictal posturing occurred in 18 of 295 (6.1%) seizures. PGES risk increased with ictal decerebration (odds ratio [OR] 14.79, 95% confidence interval [CI] 6.18-35.39, p < 0.001), decortication (OR 11.26, 95% CI 2.96-42.93, p < 0.001), or hemi-decerebration (OR 48.56, 95% CI 6.07-388.78, p < 0.001). Ictal decerebration was associated with longer PGES (p = 0.011). Postictal posturing was associated with postconvulsive central apnea (PCCA) (p = 0.004), longer hypoxemia (p < 0.001), and Spo2 recovery (p = 0.035). CONCLUSIONS: Ictal brainstem semiology is associated with increased PGES risk. Ictal decerebration is associated with longer PGES. Postictal posturing is associated with a 6-fold increased risk of PCCA, longer hypoxemia, and Spo2 recovery. Peri-ictal brainstem posturing may be a surrogate biomarker for GCS severity identifiable without in-hospital monitoring. CLASSIFICATION OF EVIDENCE: This study provides Class III evidence that peri-ictal brainstem posturing is associated with the GCS with more prolonged PGES and more severe breathing dysfunction.

An Integrative Approach to Study Structural and Functional Network Connectivity in Epilepsy Using Imaging and Signal Data.

A key area of research in epilepsy neurological disorder is the characterization of epileptic networks as they form and evolve during seizure events. In this paper, we describe the development and application of an integrative workflow to analyze functional and structural connectivity measures during seizure events using stereotactic electroencephalogram (SEEG) and diffusion weighted imaging data (DWI). We computed structural connectivity measures using electrode locations involved in recording SEEG signal data as reference points to filter fiber tracts. We used a new workflow-based tool to compute functional connectivity measures based on non-linear correlation coefficient, which allows the derivation of directed graph structures to represent coupling between signal data. We applied a hierarchical clustering based network analysis method over the functional connectivity data to characterize the organization of brain network into modules using data from 27 events across 8 seizures in a patient with refractory left insula epilepsy. The visualization of hierarchical clustering values as dendrograms shows the formation of connected clusters first within each insulae followed by merging of clusters across the two insula; however, there are clear differences between the network structures and clusters formed across the 8 seizures of the patient. The analysis of structural connectivity measures showed strong connections between contacts of certain electrodes within the same brain hemisphere with higher prevalence in the perisylvian/opercular areas. The combination of imaging and signal modalities for connectivity analysis provides information about a patient-specific dynamical functional network and examines the underlying structural connections that potentially influences the properties of the epileptic network. We also performed statistical analysis of the absolute changes in correlation values across all 8 seizures during a baseline normative time period and different seizure events, which showed decreased correlation values during seizure onset; however, the changes during ictal phases were varied.

Incidence, Recurrence, and Risk Factors for Peri-ictal Central Apnea and Sudden Unexpected Death in Epilepsy.

Introduction: Peri-ictal breathing dysfunction was proposed as a potential mechanism for SUDEP. We examined the incidence and risk factors for both ictal (ICA) and post-convulsive central apnea (PCCA) and their relationship with potential seizure severity biomarkers (i. e., post-ictal generalized EEG suppression (PGES) and recurrence. Methods: Prospective, multi-center seizure monitoring study of autonomic, and breathing biomarkers of SUDEP in adults with intractable epilepsy and monitored seizures. Video EEG, thoraco-abdominal excursions, capillary oxygen saturation, and electrocardiography were analyzed. A subgroup analysis determined the incidences of recurrent ICA and PCCA in patients with ≥2 recorded seizures. We excluded status epilepticus and obscured/unavailable video. Central apnea (absence of thoracic-abdominal breathing movements) was defined as ≥1 missed breath, and ≥5 s. ICA referred to apnea preceding or occurring along with non-convulsive seizures (NCS) or apnea before generalized convulsive seizures (GCS). Results: We analyzed 558 seizures in 218 patients (130 female); 321 seizures were NCS and 237 were GCS. ICA occurred in 180/487 (36.9%) seizures in 83/192 (43.2%) patients, all with focal epilepsy. Sleep state was related to presence of ICA [RR 1.33, CI 95% (1.08-1.64), p = 0.008] whereas extratemporal epilepsy was related to lower incidence of ICA [RR 0.58, CI 95% (0.37-0.90), p = 0.015]. ICA recurred in 45/60 (75%) patients. PCCA occurred in 41/228 (18%) of GCS in 30/134 (22.4%) patients, regardless of epilepsy type. Female sex [RR 11.30, CI 95% (4.50-28.34), p < 0.001] and ICA duration [RR 1.14 CI 95% (1.05-1.25), p = 0.001] were related to PCCA presence, whereas absence of PGES was related to absence of PCCA [0.27, CI 95% (0.16-0.47), p < 0.001]. PCCA duration was longer in males [HR 1.84, CI 95% (1.06-3.19), p = 0.003]. In 9/17 (52.9%) patients, PCCA was recurrent. Conclusion: ICA incidence is almost twice the incidence of PCCA and is only seen in focal epilepsies, as opposed to PCCA, suggesting different pathophysiologies. ICA is likely to be a recurrent semiological phenomenon of cortical seizure discharge, whereas PCCA may be a reflection of brainstem dysfunction after GCS. Prolonged ICA or PCCA may, respectively, contribute to SUDEP, as evidenced by two cases we report. Further prospective cohort studies are needed to validate these hypotheses.

Postconvulsive central apnea as a biomarker for sudden unexpected death in epilepsy (SUDEP).

OBJECTIVE: To characterize peri-ictal apnea and postictal asystole in generalized convulsive seizures (GCS) of intractable epilepsy. METHODS: This was a prospective, multicenter epilepsy monitoring study of autonomic and breathing biomarkers of sudden unexpected death in epilepsy (SUDEP) in patients ≥18 years old with intractable epilepsy and monitored GCS. Video-EEG, thoracoabdominal excursions, nasal airflow, capillary oxygen saturation, and ECG were analyzed. RESULTS: We studied 148 GCS in 87 patients. Nineteen patients had generalized epilepsy; 65 had focal epilepsy; 1 had both; and the epileptogenic zone was unknown in 2. Ictal central apnea (ICA) preceded GCS in 49 of 121 (40.4%) seizures in 23 patients, all with focal epilepsy. Postconvulsive central apnea (PCCA) occurred in 31 of 140 (22.1%) seizures in 22 patients, with generalized, focal, or unknown epileptogenic zones. In 2 patients, PCCA occurred concurrently with asystole (near-SUDEP), with an incidence rate of 10.2 per 1,000 patient-years. One patient with PCCA died of probable SUDEP during follow-up, suggesting a SUDEP incidence rate 5.1 per 1,000 patient-years. No cases of laryngospasm were detected. Rhythmic muscle artifact synchronous with breathing was present in 75 of 147 seizures and related to stertorous breathing (odds ratio 3.856, 95% confidence interval 1.395-10.663, p = 0.009). CONCLUSIONS: PCCA occurred in both focal and generalized epilepsies, suggesting a different pathophysiology from ICA, which occurred only in focal epilepsy. PCCA was seen in 2 near-SUDEP cases and 1 probable SUDEP case, suggesting that this phenomenon may serve as a clinical biomarker of SUDEP. Larger studies are needed to validate this observation. Rhythmic postictal muscle artifact is suggestive of post-GCS breathing effort rather than a specific biomarker of laryngospasm.