Minimal Reservoir Computing Generates Hippocampal-Inspired Stimulation for Open-Loop Modulation of Cortical Networks...

Area of Science:

• Computational neuroscience
• Neural engineering
• Biophysics

Background:

• High-frequency deep-brain stimulation (DBS) reduces seizures but overloads tissue.
• Stimulation effects depend on frequency and temporal patterns.
• Novel methods are needed to optimize DBS efficacy and reduce tissue load.

Purpose of the Study:

• To develop a compact reservoir model for generating hippocampal-inspired event schedules.
• To test if these schedules modulate neuronal activity in vitro with fewer pulses than standard 50-Hz stimulation.
• To compare the effects of biomimetic schedules with periodic stimulation.

Main Methods:

• A 10-unit leaky reservoir model transformed cortical input recordings into CA3-inspired outputs.
• Computational validation included retrospective analysis and held-out evaluation.
• The generated schedule was delivered open-loop to rat cortical cultures on multielectrode arrays and compared to 0.16-Hz and 50-Hz stimulation.

Main Results:

• The reservoir model showed improved performance over baseline models in spectral and temporal metrics.
• Biomimetic stimulation resulted in reduced normalized firing rate (0.8458) and burst rate (0.7792) in cortical cultures.
• A trend towards reduced firing-rate slopes was observed, though not statistically significant in the pooled analysis (p=0.0865).

Conclusions:

• An irregular, low-event stimulation schedule can heterogeneously modulate cortical network activity in vitro.
• Periodic stimulation at similar mean rates did not yield significant modulation.
• Future research should explore temporal properties and closed-loop applications in epilepsy models.