Accurately measuring a person's level of stress can have a wide variety of impacts, not only for human health, but also for the perceived feeling of safety when going after daily habits, such as walking, cycling, or driving from one place to another. While there is a vast amount of research done on stress and the related physiological responses of the human body, there is no go-to method when it comes to measuring acute stress in a live setting. This work proposes an advancement of the rule-based stress detection algorithm proposed in [1], to identify moments of stress (MOS) more reliably, through an adaptation and an individualization of the rules proposed in the original paper. The proposed algorithm leverages electrodermal activity and skin temperature, both recorded by the Empatica E4 wristband, for the assessment of an individual's stress when exposed to an audible stimulus. The algorithm achieves an average recall of 81.31%, with a precision of 46.23%, and an accuracy of 92.74%, measured on 16 test subjects. The trade-off between precision and recall can be controlled by adjusting the MOS threshold that needs to be reached for a MOS to be detected.

Human-centered applications using wearable sensors in combination with machine learning have received a great deal of attention in the last couple of years. At the same time, wearable sensors have also evolved and are now able to accurately measure physiological signals and are, therefore, suitable for detecting body reactions to stress. The field of machine learning, or more precisely, deep learning, has been able to produce outstanding results. However, in order to produce these good results, large amounts of labeled data are needed, which, in the context of physiological data related to stress detection, are a great challenge to collect, as they usually require costly experiments or expert knowledge. This usually results in an imbalanced and small dataset, which makes it difficult to train a deep learning algorithm. In recent studies, this problem is tackled with data augmentation via a Generative Adversarial Network (GAN). Conditional GANs (cGAN) are particularly suitable for this as they provide the opportunity to feed auxiliary information such as a class label into the training process to generate labeled data. However, it has been found that during the training process of GANs, different problems usually occur, such as mode collapse or vanishing gradients. To tackle the problems mentioned above, we propose a Long Short-Term Memory (LSTM) network, combined with a Fully Convolutional Network (FCN) cGAN architecture, with an additional diversity term to generate synthetic physiological data, which are used to augment the training dataset to improve the performance of a binary classifier for stress detection. We evaluated the methodology on our collected physiological measurement dataset, and we were able to show that using the method, the performance of an LSTM and an FCN classifier could be improved. Further, we showed that the generated data could not be distinguished from the real data any longer.