Recognition of Stress Activation by Unobtrusive Multi Sensing Setup

It is recognized that stress conditions play an important role in the definition of individual wellness and represent a major risk factor for most non-communicable diseases. Most studies focus on evaluation response to maximal while few them reports results about detection/monitoring mild stimulations. In this study, we investigate capability some physiological signs indicators (including Heart Rate, Rate Variability, Respiratory Galvanic Skin Response) recognize moderate cognitive activation daily life settings. To achieve goal, built up unobtrusive platform collect signals from healthy volunteers (10 subjects) undergoing via Stroop Color Word Test. We integrated our dataset with data Stress Recognition Automobile Drivers dataset. Following harmonization, signal recordings both datasets were split into five-minute blocks set 12 features was extracted each block. A feature selection implemented by two complementary approaches: Sequential Forward Feature Selection (SFFS) Auto-Encoder (AE) neural networks. Finally, explored use Self-Organizing Map (SOM) provide flexible representation status. From initial have determined, SFFS analysis, 2 (median number peaks Response signals) can discriminate statuses resting ones. addition, AE experiments also support suffice recognition. showed SOM comprehensive but compact description allowing fine prototypical