Controllability of nonlinear epileptic-seizure spreading dynamics in large-scale subject-specific brain networks

• Published in: Scientific reports Vol. 15; no. 1; pp. 6467 - 37
• Main Authors: Moosavi, S. Amin, Feldman, Jordan S., Truccolo, Wilson
• Format: Journal Article
• Language: English
• Published: London Nature Publishing Group UK 22.02.2025; Nature Publishing Group; Nature Portfolio
• Subjects: 639/766/530/2801; 639/766/530/2803; Brain - diagnostic imaging; Brain - physiopathology; Convulsions & seizures; Electric Stimulation; Electrical stimuli; Electroencephalography; Epilepsy; Epilepsy - physiopathology; Excitability; Feedback; Humanities and Social Sciences; Humans; Information processing; Latency; Models, Neurological; multidisciplinary; Nerve Net - physiopathology; Nonlinear Dynamics; Phase transitions; Prevention; Science; Science (multidisciplinary); Seizures; Seizures - physiopathology; Sensorimotor system
• DOI: 10.1038/s41598-025-90632-w
• ISSN: 2045-2322

Closed-loop electrical stimulation has become an important alternative to resective surgery for control of pharmacologically-resistant focal epileptic seizures. Seizure spread across large-scale brain networks, rather than its focal onset, is what commonly leads to major disruptions in sensorimotor and cognitive processing, as well as loss-of-consciousness, one of the main impairing aspects of the disorder. Electrical stimulation, triggered by early detection of seizure onset in epileptogenic zones (EZs), has been applied to prevent spread and its subsequent effects. Here, we show how linear feedback seizure-spread controllability in subject-specific (white-matter tractography) Epileptor network models is affected by variations in brain excitability, network coupling strength, control latency and gain, and actuation targets. Feedback control can qualitatively change the nonlinear seizure dynamics, and the paths to seizure termination and spread prevention. Notably, control onset latency is a critical parameter leading to a phase transition in spread controllability. Consequently, the efficacy of EZ-only actuation is limited depending on network excitability, coupling strength, and practical latencies for detection and actuation. Additional feedback-stabilization control of theoretically-derived optimal node subsets in the network are necessary for spread prevention. Finally, we contrast our linear-feedback controllability assessment with other measures based on minimum-energy (Gramian) controllability and nonlinear pulse-perturbation approaches.