Estimation of a single motor unit's threshold and activation range, a study on patients with muscular disorders

In clinical neurophysiology, threshold tracking studies are used to evaluate the functionality of a muscle through studying the functionality of its motor units (MUs) that govern the muscle. The functionality of an MU can be quantified by estimation of its excitability properties via MU's stimulus-response curve. In this study, we aim to develop a model-based approach to estimate MU's threshold mean and its activation range as indications of MU's excitability. This is a different approach from routine strategies in neurophysiology, which are mostly subjective.

To assess the excitability of a single MU, needle electromyography examination was used to obtain the axonal activity of that MU. To improve estimation, the examination was repeated several times on individuals. Replication of experiment introduces serial correlation between observations. We account for this correlation by using a mixed-effects model. We investigate the appropriateness of classical logistic mixed-effects model and its Bayesian formulation for estimation purpose.

Both classical and Bayesian models can obtain a reliable estimation of MU's threshold. However, we found Bayesian approach to provide a better estimate of MU's activation range. Moreover, if data contain outliers both classical and Bayesian methods are vulnerable to some extent.

Compared to the classical approach, Bayesian method is more flexible in dealing with overdispersion and provides more robust estimation of MU's parameters.