Machine Learning Model for Mild Cognitive Impairment Stage Based on Gait and MRI Images

Area of Science:

• Neurology
• Gerontology
• Biomedical Engineering

Background:

• Mild cognitive impairment (MCI) signifies reduced cognitive function, increasing dementia risk.
• Gait disturbances and brain MRI structural changes correlate with cognitive decline in MCI patients.

Purpose of the Study:

• To classify mild cognitive impairment (MCI) stages using gait parameters and brain MRI data.
• To identify reliable predictors for differentiating early-stage from late-stage MCI.

Main Methods:

• Eighty MCI patients were recruited and classified into early or late stages based on MMSE z-scores.
• A machine learning model, specifically a convolutional neural network (CNN), was trained using gait and MRI data.
• Multimodal features, including gait parameters and white matter datasets, were utilized to optimize classification.

Main Results:

• The CNN model achieved high performance in distinguishing late-stage from early-stage MCI.
• Integrating multimodal features (GAIT + white matter dataset) maximized the CNN classifier's performance.
• Single support time emerged as the strongest predictor for MCI stage classification.

Conclusions:

• Machine learning incorporating gait and white matter parameters effectively differentiates between late-stage and early-stage MCI.
• This approach offers a promising tool for objective MCI staging and monitoring disease progression.