Sensitivity analysis of a model of mammalian neural membrane

The sensitivity of the strength-duration (S-D) relationship to changes in the parameters describing the sodium channel of mammalian neuronal membrane was determined by computer simulation. A space-clamped patch of neuronal membrane was modeled by a parallel nonlinear sodium conductance, linear leakage conductance, and membrane capacitance. Each parameter that governs the activation (m) and inactivation (h) variables of the sodium channel was,varied from -50% to + 50% of its default value, and for each variation a S-D relationship was generated. Individual changes in six of the eleven parameters (alpha(m)A, alpha(m)D, alpha(h)A, beta(m)A, beta(m)B, and beta(h)B) generated substantial changes in the rheobase current and chronaxie time (Tch) of the model. Changing the parameter values individually did not correct for the model's failure to generate excitation after the release from a long duration hyperpolarization (anode break excitation). Scaling a combination of five parameters (alpha(m)A, alpha(m)B, alpha(h)A, beta(m)A, and beta(h)B) by an equal amount produced a model that generated anode break excitation and increased Tch, but also decreased the amplitude of the action potential. To reproduce the amplitude of the action potential, the maximum sodium conductance and sodium Nernst potential were increased. These modifications generated a model that had S-D properties closer to experimental results, could produce anode break excitation, and reproduced the action potential amplitude.