The usability and feasibility of a self-compassion chatbot (COMPASS) for youth living with type 1 diabetes

Although it is well established that youth with type 1 Diabetes (T1D) experience high rates of distress, current clinical care is often under-resourced and unable to provide sufficient or timely psychological support. The current study was designed to evaluate the safety, usability and feasibility of 'COMPASS,' a self-compassion chatbot intervention.

Forty adolescents (aged 12-16 years) living with T1D participated in a 12-week, single-group feasibility study using a mixed methods approach. Usability was assessed via qualitative feedback and interviews, feasibility through app analytics and recruitment data, and safety by utilisation of distress risk management protocols. Psychosocial measures (including diabetes distress, strengths and resilience, self-efficacy, self-compassion, self-care behaviours and emotional well-being) were collected at baseline, 6- and 12-week follow-ups, alongside HbA1c at baseline and 12 weeks.

Of 40 participants recruited, 32 (80%) completed the study. On average, participants completed 95% of modules started. Follow-up assessments showed promising reductions in diabetes distress, and improvements in resilience, self-efficacy, self-compassion and emotional well-being. Qualitative analysis highlighted strengths in engagement and relevance of information covered. Areas for improvement included increasing decision-making support to help alleviate the felt burden of diabetes.

The COMPASS chatbot was demonstrated to be safe, feasible and acceptable to adolescents living with T1D. Future research aims to conduct a hybrid type 1 effectiveness-implementation trial that will leverage updates in artificial intelligence to increase the flexibility and tailoring of responding.