Multimodal Perception-Based Intelligent Robots for Cognitive Function Improvement in Elderly with Mild Cognitive Impairment: A Neurocognitive Intervention Study

Authors

  • Elena Peterson *

    Laboratory of Human-Machine Interaction, Stanford University, Stanford, CA 94305, USA

Keywords:

Embodied Intelligence; Human-Robot Collaboration; Cognitive Alignment; Neurocognitive Mechanisms; Inter-Brain Synchronization; Predictive Coding

Abstract

Mild Cognitive Impairment (MCI), a critical transition between normal aging and dementia, demands early non-pharmacological intervention. Embodied robots show potential in elderly care, but existing MCI interventions rely on single-modal interaction and lack real-time cognitive state-based personalization. This study developed a multimodal perception-based intelligent robot (MPIR) and explored its effect on MCI elderly via a 10-week experiment (42 participants, MPIR vs. traditional training groups). Behavioral results: MPIR group’s MoCA score improved by 10.3% (T0:22.3±1.8 to T2:24.6±1.5, p<0.001), higher than control’s 4.2% (p<0.05), with greater ADAS-Cog reduction. Neurocognitively, MPIR group had sustained PFC/hippocampus activation and increased P300 amplitude (p<0.001). Real-time cognitive load matching and positive emotions played parallel mediating roles. Findings confirm MPIR’s effectiveness, providing new means for MCI early intervention.

Downloads

Issue

Vol. 1 No. 1 (December 2025)

Section

Articles