Human-robot interaction enhanced by a Large Language Model (LLM): analysis of the acceptability and usability of a social robot in a geriatric care institution. (Preprint)

• DOI: 10.2196/preprints.76496
• Publication Date: 2025-05-01T14:55:08Z
• Authors: Lauriane Broca Li...; Sébastien Dacunha; Xavier Alameda-Pi...; Daniel Hernández ...; Sharon Gannot; Florian Gras; Nancie Gunson; Séverin Lemaignan; Michal Polic; Pinchas Tandeitnik; Francesco Tonini; Anne-Sophie Rigaud; Maribel Pino
• Keywords: Preprint; Human–robot interaction; Institution

<sec> <title>BACKGROUND</title> Socially assistive robots offer promising opportunities to support older adults in healthcare settings by enhancing communication, reducing loneliness, and promoting emotional well-being. Beyond direct patient interaction, socially may also assist professionals facilitating information delivery, relieving workload, improving the organization of care activities. However, ensuring that these systems are both acceptable usable staff remains a critical challenge, particularly dynamic hospital environments. </sec> <title>OBJECTIVE</title> This study aims evaluate acceptability usability social robot geriatric day among patients their informal caregivers identify facilitators barriers adoption settings. <title>METHODS</title> Tests were conducted between May 2023 July 2024, involving 97 participants, including (n=65) (n=32), across three experiments, called waves. The ARI robot, developed PAL Robotics, was used this study. Participants invited interact spontaneously with as long they wished, after which completed standardized questionnaires. They administered orally complemented open-comments. <title>RESULTS</title> Quantitative analysis showed significant increase scores experimental qualitative revealed similar improvements participants reported greater comfort satisfaction, integration large language model. <title>CONCLUSIONS</title> Successive upgrades driven large-language-model integration, led increases waves, caregivers. Introducing LLM enabled engage dialogues more natural, coherent, context-sensitive. These findings show value an iterative, user-centred refinement process when integrating care. <title>CLINICALTRIAL</title> approved French national ethics committee: “Comité de Protection des Personnes, CPP Ouest II, Maison la Recherche Clinique-CHU Angers” (IRB: 2021/20) compliant General Data Regulation (GDPR) (DPO: 20210114153645, AP-HP register).