pipette
ENEnglish

Rank Deficiency in Parameter Estimation for Hodgkin-Huxley Type Current Models

M. Heinkenschloss, S. Namazifard, F. Gabbiani

Preprint

En palabras de los autores

The Jacobians of nonlinear least squares residuals play a central role when simultaneously fitting parameters of Hodgkin--Huxley (HH) type models to ionic currents measured through voltage clamp experiments. Their rank deficiency is tied to parameter degeneracy, i.e., the ability to produce the same current response with different parameters, and the Jacobians' null spaces provide insight into the structure of the parameter degeneracy, as well as into which additional experiments can reduce or even remove it. This paper analyzes Jacobian rank structure across HH-type models used for H-, A-type, fast sodium, and delayed rectifier currents. The Jacobians are rank deficient to varying degrees, depending on experimental conditions. Analytical representations of the Jacobians' null spaces are provided. In all models, one source of rank deficiency is the multiplication of the peak conductance and state variables, such as activation and inactivation, inherent to the HH-type model structure. H-current models derived from either whole cell, or cell-attached patch recordings exhibit additional rank deficiencies, implying the local existence of manifolds of parameters that lead to the same current responses. For A-type, fast sodium, and delayed rectifier models there is no additional rank deficiency, although their Jacobians may exhibit characteristics associated with ill-conditioned parameter estimation problems. Descriptions of the null spaces of these Jacobians are used to prove that the rank deficiency can be reduced or even be removed by fixing certain parameters, or by adding additional experiments. The analytical results are illustrated numerically, using data from voltage clamp experiments described in the literature.

Resultado principalEl resumen no menciona limitaciones.

Apareció: jueves, 19 de noviembre. bioRxiv. Preprint, todavía sin revisión por pares.

DOI: 10.64898/2026.11.16.767153