A reduced MIMO Wiener model for recursive identification of the depth of anesthesia. (11th November 2013)

• Record Type: Journal Article
• Authors: Silva, Margarida M.; Wigren, Torbjörn; Mendonça, Teresa
• Is Part Of: International journal of adaptive control and signal processing. Volume 28:Number 12(2014:Dec.)
• Page Start: 1357
• Page End: 1371
• Publication Date: 2013-11-11
• Subjects: Adaptive control systems -- Periodicals; Adaptive signal processing -- Periodicals; 629.836
• DOI: 10.1002/acs.2447
• Languages: English
• ISSNs: 0890-6327

Abstract

<abstract abstract-type="main" id="acs2447-abs-0001"> <title>SUMMARY</title> <p>This paper presents a modeling and identification strategy for the depth of anesthesia using the propofol and remifentanil rates as the system inputs, and the bispectral index and state entropy measurements as the systems outputs. The standard model used for this purpose has twenty two patient‐dependent parameters. This high number of parameters, the little input excitation, and the small amount of output data make classical system identification approaches unsuccessful. To overcome these issues, the new model presented in this paper has six parameters, thereby meeting the parsimony principle of system identification. An extended Kalman filter algorithm is also developed and applied to real data. The good fitting results, combined with noise suppression and a recursive update of the model parameters, are promising for the design of the depth of anesthesia controllers to be used in real time platforms. Copyright © 2013 John Wiley &amp; Sons, Ltd.</p> </abstract>