A multimodal predictive model identified acute stress with 85.37% accuracy, revealing shorter inspiration duration as a key physiological feature that is mitigated by deep breathing.
144 healthy adults (aged 18 – 44 years; 63 men)
Montreal Imaging Stress Task and deep breathing interventions
Decoding acute stress response using multimodal electrophysiological signals and predictive modeling
A data-driven predictive model accurately decoded acute stress responses using electrophysiological signals, highlighting the role of respiratory dynamics and identifying distinct stress-vulnerability phenotypes.
Dysregulated stress response has been linked to multiple health issues, yet the brain and body dynamics of acute stress remain poorly understood. Developing interpretable models is imperative in understanding this complex interplay. The present study sought to decode acute stress response by using a systematic data-driven and experimental approach to provide insights into its underlying mechanisms, with implications for accurate prediction, precision intervention, and trauma-related phenotyping. A total of 144 healthy adults (aged 18 – 44 years; 63 men) completed the Montreal Imaging Stress Task. Multimodal electrophysiological signals were recorded throughout the experiment. The interpretable predictive model achieved an excellent test accuracy of 85.37% and revealed several key stress-related features, such as shorter inspiration duration. Causal discovery analyses revealed stress-induced disruption of bottom-up inspiratory influences on frontal neural activity. Deep breathing interventions were found to increase inspiration duration and reduce perceived stress levels. The cluster analysis identified three phenotypes. The stress-vulnerable phenotype, in particular, was found to be more susceptible to stress and have greater exposure to lifetime trauma. These findings advance our understanding of the brain and body psychophysiology causal dynamics of acute stress, identify actionable targets for precision interventions, and reveal distinct phenotypes that may inform individualised stress profiles.
Building similarity graph...
Analyzing shared references across papers
Loading...
Kar Fye Alvin Lee
L. B. Liang
Theparambil A. Suhail
Neurobiology of Stress
University of Hong Kong
Hong Kong Jockey Club
Building similarity graph...
Analyzing shared references across papers
Loading...
Lee et al. (Sun,) reported a other. A multimodal predictive model identified acute stress with 85.37% accuracy, revealing shorter inspiration duration as a key physiological feature that is mitigated by deep breathing.
www.synapsesocial.com/papers/69c4cd5afdc3bde448919830 — DOI: https://doi.org/10.1016/j.ynstr.2026.100809