Quantifying Excitation-Inhibition Harmony and Its Breakdown during Epileptogenesis in a Spiking Neural Network of the Hippocampal CA1

Abstract

Epileptogenesis is often linked to disruptions in excitation-inhibition (E-I) balance, yet its quantitative characterization remains limited. Understanding and quantifying this E-I balance could enable earlier diagnosis and targeted treatment of epilepsy. We investigate competitive E-I harmony and its breakdown using a biologically grounded spiking neural network model of hippocampal CA1 that incorporates two structural pathological mechanisms: pyramidal cell (PC) axonal sprouting and loss of inhibitory interneurons. As recurrent excitatory connectivity increases via PC axonal sprouting or inhibitory interneuron populations decline through interneuron loss, the network undergoes sequential transitions from a healthy phase (HP) to an interictal pathological phase (IIP) and finally to an ictal pathological phase (IP). At the input level, we introduce the competition degree, Cd (= SE / SI), as the ratio of the strengths of excitatory (SE) to inhibitory (SI) synaptic inputs to PCs, to quantify their competitive E-I harmony. As an input-side indicator, Cd remains near 2.6 in HP, exhibits a discontinuous, abrupt jump to ~ 60 at the HP [->] IIP transition and rises further thereafter. At the output level, the corresponding output-side measure, the population-averaged mean firing rate (MFR) of PCs, , also exhibits a discontinuous, abrupt jump from ~ 4 to ~ 200 Hz at the HP [->] IIP transition and rises further thereafter, indicating progressive pathological hyperexcitability. Together, these results establish a cause-and-effect framework linking the input-side indicator Cd (cause) to the output-side measure (effect). Accordingly, Cd may serve as a potential quantitative biomarker of competitive E-I harmony and its breakdown, providing a computational causal framework that links structural pathological mechanisms to epileptogenic state transitions and offering potential for translation into diagnostic tools and therapeutic applications.