Exploring the Use of Generative AI Technologies to Create Life-Review Summaries with Family Members to Promote Mental Health in Patients with Geriatric Depression

Geriatric depression is common yet often overlooked, and existing care often fails to address daily mood fluctuations or involve companions. While reminiscence and life review can improve well-being, traditional interventions rely on facilitators, require preparation, and rarely support shared reflection. We conducted a pilot technology probe using conversational agents prototyped with Google Gemini Gems to facilitate discussions around life-review scenarios. This enabled patients and their companions to reflect, during follow-up interviews, on how generative AI could support mental health in geriatric depression through the independent and collaborative creation of life-review summaries. Each agent focused on one autobiographical element (e.g., people, events, time, place, or artifacts) and generated multi-format summaries. We then interviewed two patient-companion dyads after they experienced the generated materials. Our analysis highlights opportunities for vivid GenAI content to support mood and communication, along with concerns about usability barriers and potential emotional backfire from distressing memories. We conclude with design implications for lightweight, companion-supported life-review tools with gentle positive guidance.