Vivibot: Evaluating a Chatbot Based Positive Psychology Intervention to Promote Well-Being in Young Cancer Survivors

Introduction: Cancer is the leading cause of disease-related death in adolescents and young adults (AYA) in the United States (National Cancer Institute, 2018). AYAs with cancer are most likely to experience depression, anxiety and distress in the first 12-24 months after completing treatment (Kwak et al., 2013), making this a critical time for support. Positive psychology interventions show promise to reduce psychosocial distress associated with health adversity, and have the potential to be widely disseminated through technology (Moskowitz et al., 2012; Cohn et al., 2014). This study examines the effects of positive psychology skills delivered via the Vivibot chatbot on key psychosocial well-being outcomes in AYA cancer survivors.

Methods: AYAs (age 18 to 29) within 5 years of completing active cancer treatment were recruited within the Vivibot chatbot on Facebook messenger. Participants were randomized to either immediate access to Vivibot content (treatment) or a delayed access (control). Created using a human-centered design process with AYA cancer survivors, Vivibot content includes 4 weeks of seven positive psychology skills, daily mood ratings, video and other content produced by survivors. Control participants received only daily mood ratings for the first 4 weeks. Psychosocial well-being was assessed via online surveys at baseline, and weeks 2, 4, and 8. Mixed effects linear models were used to compare changes over the first 4 weeks between experimental groups on anxiety, depression, positive and negative emotion. Within the treatment group, regression analysis was used to examine the relationship between chatbot use and psychological outcomes.

Results: Thirty-two AYAs (23 female, average age 24.7 +- 2.9) enrolled and were randomly assigned to treatment (N=17) or control (N=15). Overall survey response rates were 66% and 69% at weeks 2 and 4, respectively. Preliminary results from this ongoing study show that after 4 weeks of Vivibot use participants in the treatment group reported an average reduction in anxiety of 1.16 standardized t-score and the control group increased reported anxiety by 1.6. Mixed effects model revealed a trend level (p=0.11) interaction between experimental group and time with an effect size of -0.47. Additionally, those in the experimental group experienced greater reductions in anxiety when they completed more lessons (R-squared=0.12, p=0.14). There were no significant (or trend level) effects by group on changes in depression, positive emotion or negative emotion.

Conclusions: Preliminary analyses show that positive psychology skills delivered via chatbot support anxiety reduction among young cancer survivors, further analyses is required to confirm the consistent pattern when the data collection is complete.