Use this URL to cite or link to this record in EThOS:
Title: A Bayesian approach to modeling excitability and experimental design of binary response single motor units
Author: Azadi , Nammam Ali
Awarding Body: Lancaster University
Current Institution: Lancaster University
Date of Award: 2011
Availability of Full Text:
Access from EThOS:
Threshold tacking studies involve studying the excitability of a motor unit (MU), the fundamental physiological unit of our motor system. The excitability property of a MU can be parameterised by its threshold and a range of stimulus intensities over which a MU shows a stochastic behaviour. Statistical methods are seldom used for this purpose and judgments are subjective. Two main interests in threshold tracking studies are i) A novel approach that makes it possible to obtain reliable estimation of the MU's excitability parameters, and ii) The efficient use of the available resources. The aim is to cover both goals. The efficiency of both classical and Bayesian approaches for estimation MU's parameters is discussed. A MU exhibits binary responses to given stimulus voltages. This makes the generalised linear models (GLMs) a standard approach to model its behaviour. The standard issue with experimental design for GLMs is the dependency of the design to the unknown model parameters. The need for optimal initial design points is another challenge faces the experimenters. We develop a sequential Monte Carlo (SMC) approach for Bayesian analysis of experimental design for binary response data that selects design point based on an online manne via minimising an expected loss. We will apply this loss function to the estimates of target quantiles from the stimulus-response curve. Through simulation we show our approach is more efficient than the existing sequential design approaches; Wu's improved Robins-Monro, generalised polya urn (GPU) based models, and 2-point design that uses D-optimality criterion. By reducing up to more than halve the length of the threshold tracking experiments, our approach uses priors on unknown parameters to tackle the dependency to the design to unknown parameters. Regardless of how to initialise the our sequential approach, its efficiency is also independent of initial design points.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available