Neural Stress Mapping with Machine Learning from EEG Data

In our rapidly changing world, the influence of stress on our mental health has become a pressing concern, necessitating innovative approaches for timely detection and intervention. Stress can affect mental as well as physical states. There are many disorders and harmful effects of stress, so our work is helpful to detect stress using EEG (electroencephalography), which detects the electrical activities happening inside the brain. The primary goal is to assess stress levels by collecting data from EEG sensors. By analyzing EEG data, there is a possibility to detect stress using the activity of neurons. To study the variation in the EEG signal, we can use different stimuli like audio and videos which could help to study the modulation. Music and videos have an impact on external and internal factors that can be studied by using this technique. To achieve this, a machine learning model i.e. Random Forest can be used and hence a stress detection system in real time is developed with an accuracy of 77%. A summary report is generated where the statistical measures such as mean, median, skewness and kurtosis are calculated with respect to the EEG data. Using this real time stress detection system, we can analyze an individual’s stress levels and provide assistance for their mental health which can help them be more productive and maintain a positive outlook on life.