Adaptive fuzzy iterative learning control based neurostimulation system and in-silico evaluation.

Area of Science:

• Neuroscience
• Control Engineering
• Biomedical Engineering

Background:

• Closed-loop neural stimulation is effective for epilepsy but relies on accurate neural models.
• Neural system complexity and uncertainty challenge the development of precise control models.
• Existing strategies struggle with model inaccuracies, limiting controller design.

Purpose of the Study:

• To propose an Adaptive Fuzzy Iterative Learning Control (AFILC) framework for model-free or model-uncertain closed-loop neural stimulation.
• To enhance neuromodulation efficacy by addressing challenges in neural system modeling.
• To improve the anti-interference capabilities and reduce reliance on initial controller parameters.

Main Methods:

• Iterative Learning Control (ILC) is utilized due to the periodic nature of neural stimulation and firing.
• A fuzzy optimization module is integrated to update ILC controller parameters in real-time.
• The AFILC strategy is evaluated using neural computational models and hardware-in-the-loop experiments.

Main Results:

• AFILC effectively suppresses epileptic states in simulations.
• The AFILC strategy demonstrates faster convergence and superior anti-interference ability compared to standard ILC.
• Hardware implementation shows good control performance and computational efficiency.

Conclusions:

• The proposed AFILC framework offers a robust solution for closed-loop neural stimulation in epilepsy, even without accurate neural models.
• AFILC provides significant advantages in convergence speed and robustness against interference.
• The successful hardware implementation paves the way for future clinical applications in epilepsy treatment.