![]() ![]() Such technologies may enable the continuous monitoring of stress. It is of paramount importance to develop robust methods for the rapid detection of human stress. Thus, stress is deeply detrimental to physiological health and psychological wellbeing. Chronic stress results in a weakened immune system, cancer, cardiovascular disease, depression, diabetes, and substance addiction. We demonstrated the potential of deep neural networks for developing robust, continuous, and noninvasive methods for stress detection and emotion classification, with the end goal of improving the quality of life. The networks’ performance exhibited a significant improvement over past methods that analyzed physiological signals for both binary stress detection and 3-class emotion classification. The deep multilayer perceptron neural network achieved 99.65% and 98.38% accuracy rates for binary and 3-class classification, respectively. The deep convolutional neural network achieved 99.80% and 99.55% accuracy rates for binary and 3-class classification, respectively. ![]() The networks were trained and tested on publicly available data collected in previous studies. The second task was 3-class classification for emotion classification, in which the networks differentiated between baseline, stressed, and amused states. The first task was binary classification for stress detection, in which the networks differentiated between stressed and non-stressed states. We tailored each neural network to analyze data from either the chest-worn (1D convolutional neural network) or wrist-worn (multilayer perceptron neural network) sensors. The deep neural networks analyzed physiological data collected from chest-worn and wrist-worn sensors to perform two tasks. Deep neural networks do not require hand-crafted features but instead extract features from raw data through the layers of the neural networks. To address this deficiency, we developed two deep neural networks: a 1-dimensional (1D) convolutional neural network and a multilayer perceptron neural network. Accuracy decreases if features are misidentified. A limitation of traditional machine learning algorithms is the requirement for hand-crafted features. Results, ranging between 50 and 90% accuracy, have been mixed. Researchers have attempted to detect stress by using traditional machine learning methods to analyze physiological signals. Such signals are collected from sensors that are attached to the human body. Prior research has shown that analyzing physiological signals is a reliable predictor of stress. Developing robust methods for the rapid and accurate detection of human stress is of paramount importance. Chronic stress results in cancer, cardiovascular disease, depression, and diabetes, and thus is deeply detrimental to physiological health and psychological wellbeing. Over 70% of Americans regularly experience stress. ![]()
0 Comments
Leave a Reply. |