Application of Heart Rate Variability (HRV) and Cortisol measurements in Applied Engineering Design Experiments

Abstract

Through interaction design experiment on human-computer interaction, the Heart Rate Variability analysis and salivary cortisol changes are tested as a marker for arousal and valence. The IDE named Mockpit 1.0 was performed in a prototype of a Ship Bridge Simulator in order to also contribute to ongoing project in TrollLabs, assessing issues on Ship Bridge design. 40 participants joined the experiment, where 32 were mechanical engineering students and 8 were employees at NTNU. There were two main tasks in the experiment, Cruise and Race, where Race was designed with intention to induce highest mental demand. In this thesis method for calculating HRV are considered and computed for 1 minute segments. The salivary cortisol was collected by Salivette with additives, and analyzed at Hormone Laboratory at Oslo University Hospital. Self-report from the participants were used as ground truth about the participants affective state, where they reported subjective arousal level and valence level. The mean HR from the same 27 selected participants showed that mean HR during last minute of Race was higher than during last minute of Cruise (MD=5.70370, SD=5.29532, sig=0.000 (2-tailed)), and that mean scores from self-reported arousal showed significant increase after task Race (MD=2.51852, SD=1.67264, sig=0.000 (2-tailed)) and the after task Race the self-reported pleasure scores decreased significant (MD= -1.74074, SD=1.67774, sig=0.000(2-tailed)). For the HRV-HF parameters, the mean difference in the last minutes of tasks was significant for all but the Welch method. The Spearman correlation study found positive correlation for valence and HRV-HF for positive emotions. Evaluations were made concerning salivary cortisol measurements, and no significant change was detected after task Race or Cruise. It is suggested that the design of this IDE is not suitable for repeated measures of salivary cortisol measurements, because recovery time after onset stimuli requires a longer relaxation time between tasks than the 5 minutes in the Mockpit 1.0 .