Skip to the content.

Neural Network for Modelling Ion Channels in Smooth Muscle Electrophysiology

Authors: Chitaranjan Mahapatra

Presentation type: Poster

Abstract

Ion channels in all types of excitable tissues play vital roles in modulating cellular electrical activities. Mathematical interpretations of these ion channels are widely used to establish a computational whole-cell model of cellular electrophysiology to investigate all these electrical activities for various pathological situations. In computational neuroscience, conventional Hodgkin-Huxley or Markov state formalisms are typically adopted to express ion channel kinetics for biophysically constrained modeling in neural and cardiac tissues. However, the interpretations of the underlying conformational changes of the ion channels are abstracted due to the assumptions of the number of transitions among the conformational states. It leads to creating model discrepancy and inaccuracy for the whole cell investigations. We have tried to reduce the assumption-based errors by using a biophysically inspired neural network differential equation model to model ion channel kinetics for the urinary bladder smooth muscle for which we have already established an electrophysiological model. We have picked up one voltage-gated potassium ion channel to provide an alternative modeling approach that could overcome certain limitations of the traditional Hodgkin-Huxley approach we initially used. Multiple ways of neural network differential equation models to add hidden states are discussed and compared. We have also elaborated on how these additional states can be helpful to reduce the missing ion channel kinetics dynamics. We have also described the limitation and additional complexities of using this neural network differential equation model.