Development of an Intelligent System for Detection of Chronic Stress from Biological Signal Processing

Our objective in this work was to develop an intelligent system able to perform an automated detection of chronic stress, based on biological signals processing and features extraction, confronted with Hans Selye clinical model of stress phase’s assessment. We recorded biological signals of blood pressure, skin surface temperature, galvanic skin resistance and heart rate of 120 health adult volunteers. Also the heart rate variability parameters were extracted for the time-domain and frequency-domain analysis. A Multi-Layer Perceptron Artificial Neural Network with a supervised learning approach was applied to generate the mathematical model. The system classification produced a precision index of 89% for identifying the stressed class, when applying the skin temperature and the heart rate variability parameters as input of the neural network, providing a satisfactory initial performance in the discrimination of stressed individuals. A worth training set with more examples mighty potentially increases the precision for identifying the stressed individuals. Despite study limitations, we consider that the use of intelligent systems to classify biological signals and identify the long term stress presence in human organism could contribute to a more objective analysis of such physiologic characteristics associated to the chronic stress and its implications in human health and performance. Future works should include different settle of physiological parameters and different machine learning techniques for analysis and classification of signals.