The self-connected group lasso is used to estimate sparse multivariable autoregressive with exogenous (MVARX) input models of the cortical interactions excited by direct current stimulation of the cortex. The group lasso criterion introduces a direct network connection between two sites only if the presence of the connection significantly reduces the mean-squared error of the model. This method is applied to intracranial recordings of the human brain to direct electrical stimulation. Excellent agreement between measured and model-predicted average responses across all data sets is obtained. One-step prediction of the recordings is also used to demonstrate that the model describes the dynamics in individual responses. We study the similarity of network models for a given set of channels when the electrical stimulation is applied at different locations in both wakefulness and nonrapid eye movement (NREM) sleep to identify common network characteristics.
Sparse multivariate autoregressive models with exogenous inputs for modeling intracerebral responses to direct electrical stimulation of the human brain / J.-. Chang, A. Pigorini, F. Seregni, M. Massimini, L. Nobili, B. Van Veen (CONFERENCE RECORD - ASILOMAR CONFERENCE ON SIGNALS, SYSTEMS, & COMPUTERS). - In: 2013 Asilomar Conference on Signals, Systems and Computers / [a cura di] M.B. Matthews. - [s.l] : IEEE, 2013. - ISBN 9781479923908. - pp. 803-807 (( Intervento presentato al 47. convegno Asilomar Conference on Signals, Systems and Computers tenutosi a Pacific Grove nel 2013.
Sparse multivariate autoregressive models with exogenous inputs for modeling intracerebral responses to direct electrical stimulation of the human brain
A. Pigorini;M. Massimini;
2013
Abstract
The self-connected group lasso is used to estimate sparse multivariable autoregressive with exogenous (MVARX) input models of the cortical interactions excited by direct current stimulation of the cortex. The group lasso criterion introduces a direct network connection between two sites only if the presence of the connection significantly reduces the mean-squared error of the model. This method is applied to intracranial recordings of the human brain to direct electrical stimulation. Excellent agreement between measured and model-predicted average responses across all data sets is obtained. One-step prediction of the recordings is also used to demonstrate that the model describes the dynamics in individual responses. We study the similarity of network models for a given set of channels when the electrical stimulation is applied at different locations in both wakefulness and nonrapid eye movement (NREM) sleep to identify common network characteristics.File | Dimensione | Formato | |
---|---|---|---|
06810397.pdf
accesso riservato
Tipologia:
Publisher's version/PDF
Dimensione
167.49 kB
Formato
Adobe PDF
|
167.49 kB | Adobe PDF | Visualizza/Apri Richiedi una copia |
Pubblicazioni consigliate
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.