Among 12 people living with dementia (PLwD) and 14 carers recruited across two focus groups, qualitative interviews revealed broadly positive attitudes toward generative AI supporting Structured Medication Reviews (SMRs), contingent on rigorous validation, transparency about AI use, and preservation of human clinical oversight. Carers emerged as critical—yet underrecognized—actors in daily polypharmacy management, monitoring side effects and supporting adherence. Empathy and personalization were identified as the core qualities making SMRs meaningful, with AI seen as a tool to reduce administrative burden rather than replace relational care.
This preprint, not yet peer-reviewed, arrives at an important intersection: dementia patients are among the highest-risk populations for polypharmacy harms—studies consistently link five or more concurrent medications to falls, cognitive decline acceleration, and hospitalizations in this group. The finding that even cognitively vulnerable patients can articulate nuanced, conditional AI acceptance challenges paternalistic assumptions that often exclude this cohort from technology design conversations. However, the sample is small (26 participants total) and UK-centric, limiting generalizability. Qualitative thematic analysis captures attitudes but cannot measure whether AI-assisted SMRs actually improve outcomes. The carer-as-safety-net dynamic identified here deserves quantitative follow-up—caregiver burden is itself a longevity risk factor. Editorially, this is confirmatory rather than paradigm-shifting work, but it meaningfully legitimizes patient-inclusive co-design frameworks for healthcare AI, a field that frequently skips end-user validation entirely.
