Using Machine Learning Techniques to Analyze Stress Levels in Electroencephalogram Data

Research in mental health has proposed the application of technologies that allow the analysis of substantial amounts of data and the provisioning of mechanisms to monitor and diagnose the health status of individuals with the objective of improving their quality of life. Mental stress has been widely addressed in this type of research as it is widely acknowledged as a condition that leads to negative impacts on modern life. Several studies have been directed to the generation of datasets with data from physiological signals obtained through sensors and the public availability of such datasets for further research. One of the main approaches for the processing and analysis of these datasets has been the use of Machine Learning techniques. This study explores the use of Machine Learning techniques to analyze a dataset consisting of mental stress level classification data from electroencephalograms. Different Machine Learning models were compared. The Multilayer Perceptron model presented the best performance with an accuracy rate of 98.99% in the predictions. The results demonstrate the potential of Machine Learning techniques as aiding tools to health monitoring and diagnosis.