We developed an excitatory-inhibitory (E-I) population model in MATLAB based on Wilson–Cowan dynamics. A single parameter, the Spine Efficacy Index (SEI), scaled excitatory synaptic gain to represent changes in dendritic spine morphology/density. Simulated local field potentials (LFPs) were derived from excitatory and inhibitory drives. We ran 20-second simulations across SEI values (0.6-1.6) with multiple trials and computed spectral power (3-80 Hz), line length, kurtosis, and synchrony (order parameter R) as biomarkers of seizure-like events (SLEs). Rescue experiments tested partial reduction of SEI or enhancement of inhibitory gain.

Increasing SEI produced a sharp transition from stable network firing to synchronized, high-amplitude oscillations with elevated theta/gamma power, increased line length, and higher synchrony (R > 0.8). A classifier using spectral and statistical features distinguished low- vs high-SEI regimes with AUC > 0.85. Rescue interventions (SEI reduction by 10-20% or inhibitory gain increase by 5-10%) normalized LFP signatures and reduced SLE occurrence. Findings were robust to ±20% variation in parameters and noise levels.

A single microstructural surrogate for dendritic spine pathology can reproduce seizure-like EEG rhythms in a computational model. This framework provides a mechanistic bridge between synaptic abnormalities and epileptiform activity and suggests spine- or inhibition-targeted rescue strategies as potential therapeutic avenues.