Volume 107, Number 6, September 2014
|Number of page(s)||6|
|Section||Interdisciplinary Physics and Related Areas of Science and Technology|
|Published online||22 September 2014|
Reconstructing neural dynamics using data assimilation with multiple models
George Mason University - Fairfax, VA, USA
Received: 29 July 2014
Accepted: 5 September 2014
Assimilation of data with models of physical processes is a critical component of modern scientific analysis. In recent years, nonlinear versions of Kalman filtering have been developed, in addition to methods that estimate model parameters in parallel with the system state. We propose a substantial extension of these tools to deal with the specific case of unmodeled variables, when training data from the variable is avaiable. The method uses a stack of several, nonidentical copies of a physical model to jointly reconstruct the variable in question. We demonstrate the ability of this technique to accurately recover an unmodeled experimental quantity, such as an ion concentration, from a single voltage trace after the training period is completed. The method is applied to reconstruct the potassium concentration in a neural culture from multielectrode array voltage measurements.
PACS: 87.19.L- – Neuroscience / 05.45.Tp – Time series analysis / 07.05.Kf – Data analysis: algorithms and implementation; data management
© EPLA, 2014
Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.
Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.
Initial download of the metrics may take a while.