By Topic

A hybrid systems model for supervisory cognitive state identification and estimation in neural prosthetics

Sign In

Cookies must be enabled to login.After enabling cookies , please use refresh or reload or ctrl+f5 on the browser for the login options.

Formats Non-Member Member
$31 $13
Learn how you can qualify for the best price for this item!
Become an IEEE Member or Subscribe to
IEEE Xplore for exclusive pricing!
close button

puzzle piece

IEEE membership options for an individual and IEEE Xplore subscriptions for an organization offer the most affordable access to essential journal articles, conference papers, standards, eBooks, and eLearning courses.

Learn more about:

IEEE membership

IEEE Xplore subscriptions

2 Author(s)
Hudson, N. ; Mech. Eng., California Inst. of Technol., Pasadena, CA, USA ; Burdick, J.W.

This paper presents a method to identify a class of hybrid system models that arise in cognitive neural prosthetic medical devices that aim to help the severely handicapped. In such systems a ¿supervisory decoder¿ is required to classify the activity of multi-unit extracellular neural recordings into a discrete set of modes that model the evolution of the brain¿s planning process. We introduce a Gibbs sampling method to identify the key parameters of a GLHMM, a hybrid dynamical system that combines a set of generalized linear models (GLM) for dynamics of neuronal signals with a hidden Markov model (HMM) that describes the discrete transitions between the brain¿s cognitive or planning states. Multiple neural signals of mixed type, including local field potentials and spike arrival times, are integrated into the model using the GLM framework. The identified model can then be used as the basis for the supervisory decoding (or estimation) of the current cognitive or planning state. The identification algorithm is applied to extracellular neural recordings obtained from set of electrodes acutely implanted in the posterior parietal cortex of a rhesus monkey. The results demonstrate the ability to accurately decode changes in behavioral or cognitive state during reaching tasks, even when the model parameters are identified from small data sets. The GLHMM models and the associated identification methods are generally applicable beyond the neural application domain.

Published in:

Decision and Control, 2008. CDC 2008. 47th IEEE Conference on

Date of Conference:

9-11 Dec. 2008