Neuro-Fuzzy Artificial Intelligent Modeling for Cognitive State Classification Using EEG Signal Analysis

Cognitive Sensing is an essential process for understanding people’s mental states, which can be classified as normal, above normal, or below normal. It requires the user’s electroencephalogram (EEG) signal data to observe the root cause of dynamic variations that drastically change the typical physiological signal patterns of individuals. This process is crucial for providing the right music therapy treatment after identifying the correct mental state of the person. Accordingly, in this paper, we propose a novel neuro-fuzzy artificial intelligent (NF-AI) model that can provide valuable logic to observe the dynamic variations from the characteristic behavior of the user’s EEG signal. This model uses a Muse EEG headband containing four dry electrodes to collect EEG signals from the significant scalp locations (TP9, AF7, AF8, and TP10) to extract brain activity patterns associated with different mental states, such as happiness, sadness, anger, and fear, from which solid features are obtained. These features are utilized to train a pre-trained deep neural network (DNN) for providing better classification outcomes about the person’s mental state, categorizing it as normal, above normal, or below normal. Our NF-AI model shows promising potential for the cognitive sensing of mental states, offering a robust tool for the selection of personalized music therapy treatments.