Mental Workload Detection And Assessment Using Brain Signals

Abstract

Electroencephalogram (EEG) signals can be used to identify mental workload, which is a difficult topic with important industrial and medical implications. In this article, we offer a method for identifying a person's mental workload from EEG signals by utilizing deep learning models and signal processing methods. We gathered a collection of EEG signals from participants in cognitive tasks with varied degrees of difficulty. Using independent component analysis (ICA) and bandpass filters, we preprocessed the data to remove noise and artifacts. Additionally, we extracted custom characteristics, such as time domain, frequency domain, and entropy domain features. The performance of deep learning models including recurrent neural networks (RNN) and long short-term memory (LSTM) networks were compared, and the accuracy of the model was optimized, according to our analysis report. In order to assess the model's efficacy over the dataset we gave, we used automatic feature extraction methods such as the wavelet transform, independent component analysis (ICA), and time-frequency analysis. Our results showed that machine learning achieved an accuracy of 83% in classifying mental workload levels. The model's performance was further improved by using automatic feature extraction techniques, resulting in an accuracy of 88.3%.