Authors :
Gagan Acharya, PhD Candidate – University of California, Riverside
Andrew Huang, PhD Candidate – University of California Riverside
Viji Santhakumar, PhD – University of California, Riverside
Presenting Author: Erfan Nozari, PhD – University of California, Riverside
Rationale:
Despite advances in neurostimulation for treating neurological and psychiatric conditions, outcomes for epilepsy remain suboptimal. FDA-approved neuromodulation devices for drug-resistant epilepsy attempt to abort seizures by delivering high-frequency electrical pulses into brain regions that are already hyperexcitable or dynamically unstable. While this strategy can occasionally desynchronize or suppress neural populations or induce long-term plasticity, clinical results are limited—only about 20% of patients achieve long-term seizure freedom. This calls for new paradigms that are more physiologically aligned with the energy dynamics of seizure generation.
Methods:
We propose a passivity-based control (PBC) framework for seizure suppression that fundamentally departs from conventional stimulation by withdrawing energy from epileptic circuits rather than injecting it. Rooted in well-established principles from control theory, PBC has been widely used in engineering to stabilize nonlinear systems but remains virtually unexplored in neuromodulation. We integrate PBC with a real-time, model-driven seizure detection module to construct a closed-loop, continuous-time control system for seizure intervention. To evaluate the system’s efficacy, we apply it to a biophysically realistic dentate gyrus network model (Santhakumar et al., 2005) capable of producing both evoked and spontaneous seizure-like discharges. The model incorporates detailed cell types and network connectivity, enabling us to test the control system under biologically plausible dynamics.
Results:
Simulations demonstrate that PBC is highly effective at terminating seizure-like activity with minimal stimulation. In evoked seizure scenarios, applying PBC just 20 ms after seizure onset fully suppressed ictal spiking, even when control onset was delayed by up to 30 ms—resulting in a 94% average reduction in spiking activity (Wilcoxon signed-rank test, p < 0.01). In spontaneous seizure scenarios, where events emerged stochastically across the network, PBC reduced total seizure duration by 93% and ictal spiking by 92%, both statistically significant (p < 0.01, Wilcoxon signed-rank test).