This research introduces a novel simulation framework that combines reinforcement learning, large language models, and clinical expertise to improve robotic assistance for dementia patients.

• Addresses data limitations in assistive robotics through realistic simulated environments
• Models cognitive and emotional states of persons living with dementia
• Enables robots to adapt their interaction strategies in real-time
• Creates a testbed for safe optimization before real-world deployment

This approach has significant implications for healthcare delivery, potentially enhancing quality of life for dementia patients while reducing caregiver burden in an aging population where care resources are increasingly strained.

Integrating Reinforcement Learning and AI Agents for Adaptive Robotic Interaction and Assistance in Dementia Care