Robust Hidden Markov Model Selection by Bayesian Network-Based Algorithms Embracing Various Problems of Experimental Data

Abstract

Listen

Translate article

We present results of a Bayesian network analysis method for posterior estimation of the rates of a linear stochastic network for ion channel data. The Bayesian network is an analytical stochastic approximation of the underlying master equation, enabling a precise modeling of stochastic processes that are traditionally modeled by deterministic rate equations. The full Bayesian treatment of macroscopic patch clamp (PC) and related fluorescence data has superior model selection and inference properties over traditional maximum likelihood techniques and deterministic rate equations.