AI-driven assessment of a social media intervention for AD/ADRD caregivers: a stage II-III trial of 966 participants

Digital community support for family members of a person with Alzheimer’s disease and related dementias (AD/ADRD) is promising, yet its effectiveness is unknown. In an NIH Stage II/III clinical trial ( n = 966), African American and Hispanic family caregivers, aged 18 and older, received daily social support messages for 12 months. The primary outcomes were the emotional valence score, quantified using natural language processing, and the proportion of isolated participants who were continuously captured on the platform. High retention was achieved, with 82.3% of participants completing ≥ 9 months despite 86.7% of the cohort having low digital literacy. Both groups showed significant improvements in emotional valence scores ( p < 0.001), resulting in moderate-to-large effect sizes (Cohen’s d = 0.756). Furthermore, both groups demonstrated a large effect size in reducing the number of isolates, with a 46.2% reduction (Cohen’s h = 1.144). This trial demonstrates that AI-monitored, social media-based interventions can effectively improve emotional well-being and reduce the number of isolates among African American and Hispanic family caregivers. The high retention and moderate-to-large effect sizes suggest that leveraging common, accessible digital platforms can successfully overcome traditional barriers to care and clinical trial participation for digitally underserved populations.