Fractional-order model predictive control as a framework for electrical neurostimulation in epilepsy

Area of Science:

• Neuroscience and Biomedical Engineering
• Control Systems Theory
• Computational Medicine

Background:

• Current electrical neurostimulation for epilepsy relies on limited sensing and pre-set stimulation doses, underutilizing sensor data.
• Existing event-triggered strategies lack theoretical guarantees and optimal performance for neurological conditions.
• Therapeutic neurostimulation requires adaptive, real-time feedback mechanisms to improve efficacy.

Purpose of the Study:

• To develop a model-based, real-time closed-loop electrical neurostimulation strategy.
• To leverage fractional-order systems (FOS) for improved modeling of biological system dynamics.
• To enhance epileptic seizure mitigation through adaptive stimulation control.

Main Methods:

• Proposed a model predictive control (MPC) framework with an underlying fractional-order system (FOS) predictive model.
• Implemented a real-time closed-loop strategy adapting stimulation based on neurophysiological state estimation.
• Validated the approach through computational simulations of seizure-like events and analysis of real epilepsy data.

Main Results:

• Demonstrated the effectiveness of the FOS-based MPC approach in mitigating simulated epileptic seizures.
• Showcased the framework's ability to capture long-term dependencies in biological systems.
• Validated the method using established neuroscience models and real patient seizure data.

Conclusions:

• The proposed model-based closed-loop neurostimulation framework offers robust, real-time adaptive control.
• Fractional-order dynamics and MPC provide a powerful combination for advanced neurostimulation strategies.
• This study paves the way for developing more effective closed-loop devices for epilepsy treatment.