Comparison of FIR and ANFIS methodologies for prediction of mean blood pressure and auditory evoked potentials index during anaesthesia

Abstract

During anaesthesia mean blood pressure (MBP) is monitored to maintain haemodynamic stability and to assess the level of consciousness. Auditory Evoked Potentials (AEP) are monitored. The purpose of this paper is to compare two soft computing methodologies in terms of prediction of MBP and an AEP-index (AEPi). The Fuzzy Inductive Reasoning (FIR) is a methodology derived from the General System Theory that allows to study the conceptual behaviour modes of systems. The main tasks of FIR are the identification of qualitative models and the prediction of future output states. The Adaptive-Network-based Fuzzy Inference System (ANFIS) is a hybrid neuro-fuzzy methodology, i.e. a fuzzy inference system (FIS) tuned with backpropagation algorithm based on input-output pairs comprising the training data. The FIR model identification technique was used to obtain the causal and temporal structure (relevant inputs and delays) of the models that represent the systems under study. These structures were used by both ANFIS and FIR for the prediction of future output states. The results showed that both methodologies were able to predict MBP and DAI; however no significant differences between the methodologies were found.