Enhancing Mild Cognitive Impairment Auxiliary Identification Through Multimodal Cognitive Assessment with Eye...

Area of Science:

• Neuroscience
• Artificial Intelligence
• Biomedical Engineering

Background:

• Mild Cognitive Impairment (MCI) represents a crucial stage between normal aging and dementia, necessitating early detection to prevent cognitive decline.
• Conventional cognitive assessments like MMSE and MoCA have limitations in early-stage MCI detection.
• This research introduces a novel supportive system for MCI identification using eye-tracking and convolutional neural network (CNN) analysis.

Purpose of the Study:

• To develop and validate a multimodal auxiliary identification system for early MCI detection.
• To evaluate the effectiveness of eye-tracking features, behavioral features, and their combination in MCI classification.
• To compare the performance of single-feature models against a combined multimodal model.

Main Methods:

• A multimodal auxiliary identification model was created using eye-tracking technology and machine learning.
• 128 participants (40 MCI, 57 elderly controls, 31 young adults) underwent four eye movement tasks and two cognitive tests.
• 31 eye movement and 8 behavioral features were extracted and analyzed using CNN for classification accuracy.

Main Results:

• The combined features model demonstrated superior discrimination accuracy compared to single-feature models.
• The multimodal model achieved an average accuracy of 74.62% in differentiating MCI from healthy individuals (including young adults).
• Distinguishing MCI from elderly controls yielded an average accuracy of 66.50% with the combined model.

Conclusions:

• Multimodal models integrating eye-tracking and behavioral data significantly outperform single-feature models for MCI identification.
• Eye-tracking technology shows substantial potential for the early detection of MCI.
• This approach offers a novel pathway for enhancing the effectiveness of community-based early detection of MCI.