Mindfulness based stress reduction: A randomised trial of a virtual human, teletherapy, and a chatbot

Psychological stress is a significant issue amongst University students. Mindfulness based stress reduction (MBSR) interventions can significantly reduce stress, in both face to face and digital formats. However, adherence to digital interventions is typically low. Virtual humans may improve adherence through their increased humanlike-ness and rapport-building behaviors. This study aimed to investigate whether there were any differences in response to MBSR when delivered by a virtual human, a human teletherapist or a chatbot. A randomized controlled trial (RCT) was conducted with 108 stressed University students randomized to one of the three delivery methods. Participants attended one in-person session of an MBSR intervention using their allocated delivery method and were asked to complete online homework at least twice a week for the following four weeks. The primary outcome was perceived stress, with secondary outcomes of physiological stress indices over the session, mindfulness, homework adherence, and perceived empathy of the agent. Results showed significant reductions in stress and increases in mindfulness across all groups over time. Peripheral skin temperature significantly increased across all three groups, but only the teletherapy group had significantly increased electrodermal activity after the in-person session. There were no significant effects on heart rate or indices of heart rate variability. Homework adherence was significantly greater in the virtual human group compared to the teletherapy group, and homework engagement was lowest in the chatbot group. The teletherapist was rated significantly more empathetic than the other agents. Overall, these results support the use of all three digital interventions for delivering MBSR. Virtual humans may have the advantage of increasing adherence to homework mindfulness sessions in student populations, but more work is needed to increase perceived empathy and to replicate these results in other populations.