MEMOR-E: In-Context and Fine-Tuned LLM Personalization for Alzheimer's Assistive Robotics
MEMOR-E is a newly introduced mobile quadruped robot designed to assist Alzheimer's patients and their caregivers, as detailed in an arXiv paper published on May 26, 2026. Equipped with an interactive tablet interface, the robot offers various support functions, including medication reminders, routine guidance, memory-oriented interactions, and companionship. The research focused on evaluating the feasibility of using large language models (LLMs) for personalized assistance, specifically by fine-tuning them to emulate stage-consistent cognitive behavior and interpret responses across standard neuropsychological language tasks. The training data for this process comprised audio transcriptions from 235 Alzheimer's patients, supplemented by synthetically generated healthy controls. Additionally, the study explored the application of in-context learning (ICL) within LLMs, where a second LLM was utilized to produce domain and severity-level cognitive error summaries, enhancing the system's analytical capabilities. The findings indicate that MEMOR-E can generate stage-aware, non-diagnostic cognitive summaries, which are crucial for tailoring assistive interactions to individual patient needs. Furthermore, the system incorporates explainable AI mechanisms to translate complex model outputs into transparent, human-readable evidence. This feature is designed to facilitate caregiver oversight and foster trust in the assistive technology, demonstrating a significant step towards practical, personalized robotic assistance in healthcare.
Developers can leverage LLMs for highly personalized, context-aware applications in sensitive domains like healthcare, integrating explainable AI for transparency and trust.