Conversational Mental Health Agents | Raju Maharjan
Image of a desktop workspace showing analaysis of Google Home data

Partners
Technical University of Denmark: Health Technology Department, University of Copenhagen: Public Health Department

Process
Literature Review, Conversational Design, Lab Study, Interviewing, In-The-Wild Deployment

Growing rates of mental illness, including depression and anxiety, are placing ever-increased pressure on health services and professionals nationally and globally. In response to these emerging public health challenges, many have turned towards technology for answers. Health technology researchers and developers alike have focused on the capacity of mobile devices in particular to support and expand access to care through practices of automated data collection, self-report and intervention.

Smartphones and smartwatches have received considerable attention in this regard for their relative low-cost, near-ubiquity, accessibility and connectivity. And yet, these technologies are increasingly but one possible medium for digital interaction, impose certain inescapable constraints on users' self-expression, and often require significant effort to support the engagement of users. At the same time, the growing popularity of commercial smartspeaker systems, from Amazon's Alexa to Google's Home, has made voice-based interaction via conversational agent an everyday reality for many. These converging developments have surfaced the possibility of employing smartspeaker devices to support the self-report, and possibly even treatment, of diverse mental health conditions; although we yet know little as to the feasibility, ethics nor experience of doing so.

The work of Dr. Raju Maharjan's PhD faced these questions head on; investigating in particular whether and how smartspeaker devices might be designed to support the conversational self-report of mental health in the at-home context. Through literature review, automated topic modelling, interview, lab and real-world deployment studies, Raju identified key factors in support of the functional, reliable, usable and pleasurable design of conversational health agents, demonstrated significant correlation between self-reports of wellbeing provided on-paper and via conversational agent, and provided rich insight into at-home smartspeaker interactions as experienced by users with depression and bipolar disorder — initial evidence for the feasibility of conversational agents as means to support the self-report of mental health and wellbeing.

Read more about Raju's work including the Sofia conversational agent he developed here.


Publications

Maharjan, R., Doherty, K., Bækgaard, P. & Bardram, J. (2022, In Press). What’s Up With Conversational Health Agents? Subjects of Criticism, Needs and Design Implications. Frontiers in Digital Health.

Maharjan, R., Doherty, K., Rohani, D., Bækgaard, P. & Bardram, J. (2022, In Press). Experiences of a Speech-Enabled Conversational Agent for the Self-Report of Wellbeing Among People Living with Affective Disorders: An In-The-Wild Study. ACM Transactions on Interactive Intelligent Systems (TiiS). ACM.

Maharjan, R., Doherty, K., Rohani, D., Bækgaard, P. & Bardram, J. (2022, In Press). What’s the Difference? Investigating the Self-Report of Wellbeing via Conversational Agent and Web App. IEEE Pervasive Computing Special Issue on Mental Health, Mood, and Emotion. IEEE.

Maharjan, R., Rohani, D., Bækgaard, P., Bardram, J. & Doherty, K. (2021, July). Can We Talk? Design Implications for the Questionnaire-Driven Self-Report of Health and Wellbeing via Conversational Agent. CUI'21. ACM.