Adaptive Neuro-Fuzzy Inference System for Cognitive Load Assessment in Brain Machine Interfaces

An Adaptive Neuro-Fuzzy Inference System (ANFIS) for assessing cognitive load in Brain-Machine Interfaces (BMIs) is suggested to be developed and evaluated in this work. EEG characteristics and task-related metrics are only two examples of the pertinent input variables that the ANFIS model uses to forecast and evaluate cognitive load levels. In order to deal with the inherent imprecision and uncertainty in cognitive load data, fuzzy learning is utilised to convert discrete input values into fuzzy sets through the use of predefined membership functions. To represent the complex interactions between input variables and cognitive load, rule activation and inference mimic the cognitive process of integrating data from several rules. The resulting fuzzy output is then defuzzed using techniques like centroid or mean of maxima to yield an understandable and straightforward measure of cognitive burden. The efficacy of the ANFIS model in offering a comprehensive and precise assessment of cognitive load levels in BMIs is demonstrated. The ANFIS model has potential as a reliable and flexible method for evaluating cognitive stress in the context of brain-machine interfaces. The study shows superior performance over an existing KNN method with an accuracy of ${9 0 \%}$, precision of 88%, recall of 82%, and sensitivity of 85%.