Neuro-fuzzy techniques to model pharmacodynamic interactions between anesthetic drugs on the bispectral index: A preliminary study

Article type: Research Article

Authors: Nunes, Catarina S. | Ferreira, David A. | Mendonça, Teresa F. | Amorim, Pedro | Antunes, Luís M.

Affiliations: Departamento de Matemática Aplicada, Faculdade de Ciências da Universidade do Porto, Portugal | CECAV, Universidade de Trás-os-Montes e Alto Douro, Portugal | Departamento de Anestesiologia, Hospital Geral de Santo António, Porto, Portugal

Note: [] Corresponding author: Prof. Catarina S. Nunes, Departamento de Matemática Aplicada, Faculdade de Ciências da Universidade do Porto, Rua do Campo Alegre 687, 4169-007 Porto, Portugal. Tel.: +35 1 2201 00882; Fax: +35 1 220 100809; E-mail: [email protected]

Abstract: The aim of this preliminary study was to try to identify the effect of propofol and remifentanil on bispectral index of the electroencephalogram (BIS), and to determine if a model can be generalized between different patients. We used the data from three patients under stable propofol and remifentanil anesthesia for neurosurgical procedures. A Takagi-Sugeno-Kang (TSK) fuzzy model was used to describe the effect of interaction of propofol and remifentanil on BIS. An Adaptive Network-Fuzzy Inference System (ANFIS) was used to obtain the parameters of the TSK model. The ANFIS was able to optimize the parameters of the fuzzy TSK model. The obtained TSK model reflected the data trend and was able to capture the synergistic interaction between propofol and remifentanil during the maintenance phase of anesthesia. The model was trained with the data of two patients and tested with the data of a third patient. Different patients have different physiological responses and this preliminary study results suggested that the BIS values were independent of patients' variability. This result should be improved using a larger set of data with more patients.

Journal: Journal of Intelligent & Fuzzy Systems, vol. 16, no. 1, pp. 15-22, 2005