Thesis titled "Intelligent Stress Detection System Based on Shimmer Sensors" is dedicated to creating an automated solution for detecting stress using biosignals. The main goal of the work was to develop a methodology and software that would allow effective monitoring of a person's physiological state and identifying stress states using data obtained from photoplethysmography and electrodermal activity sensors.
A series of experiments with 30 participants were conducted, using an air raid alarm signal as the stressor.
Methods of filtering and signal normalization were applied for data processing. An analysis of heart rate variability and electrodermal activity was performed to determine the main parameters characterizing the stress state.
Several machine learning methods were used to classify states of calm and stress, including K-Nearest Neighbors, Naive Bayes, Random Forest, Decision Tree, and Support Vector Regression. Based on the obtained results, the Naive Bayes method showed the highest classification accuracy of 88%. The developed software automates the process of data processing, analysis, and stress state classification.
The practical significance of the work lies in the possibility of using the created system to monitor a person's physiological state, allowing for the quick detection of stress states and providing recommendations or warnings to the user. This contributes to improving health and safety levels and can also be applied in medical diagnostics and the prevention of stress-related diseases.
Research advisor: Y.Kuts
