Feasibility of Using Wearable Sensors to Measure Listening-Related Stress

Purpose: This study evaluated the feasibility of using wearable sensors to assess listening-related stress in varying listening difficulties. The results were compared to listening effort assessed using laboratory-grade physiological measures. We hypothesized that these wearable sensors could accurately measure listening-related stress and that the results would be in the same direction as those observed with laboratory-grade equipment. Method: A within-subject repeated-measures study was conducted on 16 young adults with typical hearing ability. Participants wore three commercially available sensors to record heart rate (HR), respiration rate (RR), and galvanic skin response (GSR). Participants' listening-related stress was assessed using self-report, wearable sensors, and laboratory-grade physiological measures while listening to sentences in noise. Listening conditions were manipulated in terms of signal-to-noise ratios (SNRs; −4, 0, +4, and +8 dB SNR), sentence predictability (high and low), and stress-inducing negative feedback (flash of light and monetary punishers). Results: On average, participants reported higher arousal and task load with decreasing SNR, for low predictability, and with negative feedback. Participants had significantly higher HR and RR in difficult listening conditions, when measured with wearable sensors. However, no significant differences were observed for skin conductance. There was a good agreement between the measures for HR but not for RR and GSR. Conclusions: Although wearable sensors have potential in reflecting changes in listening effort, the results were not consistent with laboratory-grade measures. Methodological choices in both measures and design appear to affect outcomes. Future studies should focus on crafting designs appropriate to wearable devices to reflect changes in listening-related stress.