The intelligent architecture for simulation of inhalational anaesthesia
Journal
Biomedical Engineering - Applications, Basis and Communications
Journal Volume
16
Journal Issue
5
Date Issued
2004-10-25
Author(s)
Abstract
In order to simulate the whole operation during inhalational anaesthesia, an intelligent architecture for the monitoring, control and modelling includes four blocks, which are a hierarchical structure for monitoring depth of anaesthesia (DOA), drug controller, patient model, and vaporizer model. In the first block, using the electroencephalograph (EEG) signals (i.e., bispectral index (BIS)), end-tidal anesthetic agents (Etaa) as the first level for primary DOA, and the systolic arterial pressure (SAP) and heart rate (HR) as the second level for secondary DOA, a hierarchical structure is therefore used to merge these two levels to decide DOA via fuzzy model. Block 2 is a drug controller that controls the drug infusion to the patient using a fuzzy logic controller. In the third block, a four-input and four-output neural network has been designed for patient model. Finally, in the fourth block, a single input and output neural network has also been designed for vaporizer model. Simulating the 13 clinical patients' status with different set points of DOA (defined from 0 - 100), the average of drug consumptions of the anaesthetic gas is 0.93. ± 0.26, 0.69 ± 0.24, and 0.74 ± 0.21% when setting the set point of DOA at 50 (i.e., anaesthetic ok), 30 (i.e., anaesthetic small deep), and 10 (i.e., anaesthetic deep), respectively. In comparison with the routine standard practice group (i.e., 0.69 ± 0.30%) using the same inhalational gas, there is a significant difference in DOA at 50 (p < 0.05) but no any significant differences in DOA at 30 or 10 (p > 0.05). We conclude that the routine standard practice of clinical trials is always controlled DOA at 30 and 10 which are lesser drugs used than DOA at 50.
Subjects
Bispectral index | Depth of anaesthesia | Electroencephalograph signals | Fuzzy logic controller | Fuzzy model | Inhalational anaesthesia | Neural network
Type
journal article