MENTAL WORKLOAD DETECTION USING DEEP LEARNING ON EEG SIGNALS

This study presents a hybrid deep learning model for classifying mental workload (MWL) using electroencephalography (EEG) data.Leveraging the publicly available STEW dataset, the model integrates a Convolutional Neural Network (CNN) for spatial feature extraction with a Long Short-Term Memory (LSTM) network to capture temporal dynamics.Through a 5-fold cross-validation methodology, the system successfully classifies EEG signals into four distinct workload levels, achieving an average accuracy of 70.77% and a macroaveraged F1-score of 70.81%.The model demonstrates particular effectiveness in identifying extreme workload states (low and high), underscoring its potential for deployment in adaptive cognitive monitoring systems designed to enhance human performance and safety.