Caregiver Acceptability of an LLM-Powered Assistant Interface to Improve Sleep Quality of the Elderly

While increased longevity and improved health at older ages are notable achievements of the 21st century, they pose significant challenges, particularly in informal care. This paper discusses NightCare Assistant, a human-centric system that uses nighttime data and Large Language Models (LLM) to translate sleep quality into practical suggestions for improving the daily routine of ageing individuals. These suggestions are displayed through a tablet application, allowing caregivers to interact with an AI assistant. The research aims to understand the acceptability of this solution, focusing on caregivers’ reactions to different AI-generated outputs and their willingness to follow these suggestions. An online questionnaire tested the system’s acceptability, examining caregivers’ perceptions of text, image, and audio outputs generated by AI. Results indicate significant interest among caregivers in adopting technological solutions to ease caregiving responsibilities. Caregivers found audio feedback the most reliable and understandable, followed by text and image outputs. Caregivers’ willingness to follow NightCare Assistant’s advice supports the system’s potential to improve care quality and safety for ageing individuals. The study highlights the necessity of addressing both nighttime behaviours and daily routines to effectively support cognitive health. NightCare Assistant’s integration of nighttime monitoring data into actionable daily interventions provides a comprehensive strategy for enhancing the well-being of older adults and their caregivers. Its acceptance among caregivers suggests that similar technological interventions can support autonomous living and reduce caregiver burden.