Multiple Therapeutic Mechanisms of an Automated Self-Guided AI-Enhanced Psychoeducation Intervention for Dementia...

PDC30, an online psychoeducational intervention, was found to improve depression, burden, and positive gains in dementia family caregivers, compared with a 1-month waitlist control. However, the therapeutic mechanism is unclear. This study investigated whether caregiving self-efficacy and positive reappraisal coping mediated the intervention effects.

Longitudinal mediation within a parallel-group randomized waitlist-controlled trial.

Web-based platform accessible 24/7 globally.

From 43 countries, 274 caregivers who provided ≥10 weekly care hours to relatives with dementia, scored ≥5 on Patient Health Questionnaire-9 (PHQ-9), and were without recent intervention.

A self-guided automated intervention, PDC30 combines a Guidebook, a graphic novel (for caregivers with children), an AI-powered counseling chatbot, and interactive tools to support cognitive-behavioral strategies, relaxation practices, and activities that strengthen caregiver-recipient relationships.

The primary outcome was depression (PHQ-9), and secondary outcomes were burden and positive gains. Mediators were self-efficacy beliefs (responding to disruptive behaviors [SE-RDB] and controlling upsetting thoughts [SE-CUT]) and reappraisal coping. Following baseline measurement, mediating and outcome variables were re-assessed at 3 weeks and 1 month respectively. Missing values were replaced by multiple imputation.

Path model estimated using restricted maximum likelihood showed that improvement in depression was mediated by SE-CUT (indirect effect β = -0.033) and reappraisal coping (β = -0.043); burden reduction by SE-CUT (β = -0.037) and SE-RDB (β = -0.041); and positive gains increase by reappraisal coping (β = 0.048).

While SE-RDB was partly responsible for PDC30's effect on burden, the main drivers of the impacts on mood, positive gains, and burden, were increased reappraisal coping and SE-CUT.