Explainable stress type classification captures physiologically relevant responses in the Maastricht Acute Stress Test.

Introduction: Current stress detection methods concentrate on identification of stress and non-stress states despite the existence of various stress types. The present study performs a more specific, explainable stress classification, which could provide valuable information on the physiological stress reactions. Methods: Physiological responses were measured in the Maastricht Acute Stress Test (MAST), comprising alternating trials of cold pressor (inducing physiological stress and pain) and mental arithmetics (eliciting cognitive and social-evaluative stress). The responses in these subtasks were compared to each other and to the baseline through mixed model analysis. Subsequently, stress type detection was conducted with a comprehensive analysis of several machine learning components affecting classification. Finally, explainable artificial intelligence (XAI) methods were applied to analyze the influence of physiological features on model behavior. Results: Most of the investigated physiological reactions were specific to the stressors, and the subtasks could be distinguished from baseline with up to 86.5% balanced accuracy. The choice of the physiological signals to measure (up to 25%-point difference in balanced accuracy) and the selection of features (up to 7%-point difference) were the two key components in classification. Reflection of the XAI analysis to mixed model results and human physiology revealed that the stress detection model concentrated on physiological features relevant for the two stressors. Discussion: The findings confirm that multimodal machine learning classification can detect different types of stress reactions from baseline while focusing on physiologically sensible changes. Since the measured signals and feature selection affected classification performance the most, data analytic choices left limited input information uncompensated.

Personalized mental stress detection with self-organizing map: From laboratory to the field.

Stress has become a major health concern and there is a need to study and develop new digital means for real-time stress detection. Currently, the majority of stress detection research is using population based approaches that lack the capability to adapt to individual differences. They also use supervised learning methods, requiring extensive labeling of training data, and they are typically tested on data collected in a laboratory and thus do not generalize to field conditions. To address these issues, we present multiple personalized models based on an unsupervised algorithm, the Self-Organizing Map (SOM), and we propose an algorithmic pipeline to apply the method for both laboratory and field data. The performance is evaluated on a dataset of physiological measurements from a laboratory test and on a field dataset consisting of four weeks of physiological and smartphone usage data. In these tests, the performance on the field data was steady across the different personalization levels (accuracy around 60%) and a fully personalized model performed the best on the laboratory data, achieving accuracy of 92% which is comparable to state-of-the-art supervised classifiers. These results demonstrate the feasibility of SOM in personalized mental stress detection both in constrained and free-living environment.