Forearm Surface Electromyographic Signal Analysis for Stress Classification Using Neural Networks*
Session Number
2
Advisor(s)
Dr. Ashwin Mohan, IMSA SYNAPSE Lab
Location
A150
Discipline
Medical and Health Sciences
Start Date
15-4-2026 11:10 AM
End Date
15-4-2026 11:55 AM
Abstract
Acute stress can have mental and physiological ramifications, affecting cognitive performance and learning outcomes. This study investigated how short-term stress can be identified through forearm surface electromyographic (sEMG) signal analysis in academic and social situations. Participants underwent the Trier Social Stress Test and were exposed to contrasting high-stress and low-stress conditions involving time pressure and evaluation cues. Stress-inducing circumstances include delivering a speech and completing mental math under time constraints. Self-reported stress levels were measured before, after, and during the test, bolstering sEMG data collected by surface electrodes. The sEMG signals were collected using Backyard Brains’ Human Spikebox, reading data from flexor and extensor muscles by electrodes on the participant’s forearm. The data analysis was conducted on Matlab and stress-states were evaluated by a neural network. These findings suggest there is a link between stressed states and extracted sEMG features in the frequency domain, including amplitude, root-mean-squared values, and slope sign changes. Results indicate significant changes in sEMG signals during stress-inducing sections, with distinct features extracted in analysis. Evaluating valid stress identifiers is vital for developing stress-management strategies, enabling preventative or active measures to improve performance in academic and social environments.
Forearm Surface Electromyographic Signal Analysis for Stress Classification Using Neural Networks*
A150
Acute stress can have mental and physiological ramifications, affecting cognitive performance and learning outcomes. This study investigated how short-term stress can be identified through forearm surface electromyographic (sEMG) signal analysis in academic and social situations. Participants underwent the Trier Social Stress Test and were exposed to contrasting high-stress and low-stress conditions involving time pressure and evaluation cues. Stress-inducing circumstances include delivering a speech and completing mental math under time constraints. Self-reported stress levels were measured before, after, and during the test, bolstering sEMG data collected by surface electrodes. The sEMG signals were collected using Backyard Brains’ Human Spikebox, reading data from flexor and extensor muscles by electrodes on the participant’s forearm. The data analysis was conducted on Matlab and stress-states were evaluated by a neural network. These findings suggest there is a link between stressed states and extracted sEMG features in the frequency domain, including amplitude, root-mean-squared values, and slope sign changes. Results indicate significant changes in sEMG signals during stress-inducing sections, with distinct features extracted in analysis. Evaluating valid stress identifiers is vital for developing stress-management strategies, enabling preventative or active measures to improve performance in academic and social environments.