Mild cognitive impairment, dementia, and cognitive dysfunction screening using machine learning.

Area of Science:

• Artificial Intelligence in Medicine
• Neuroscience
• Cognitive Health

Background:

• Cognitive dysfunction, including mild cognitive impairment (MCI) and dementia, poses a significant public health challenge.
• Accurate and early screening for cognitive decline is crucial for timely intervention and management.
• Current screening methods can be enhanced with advanced analytical tools.

Purpose of the Study:

• To develop and evaluate a machine learning (ML) algorithm for identifying cognitive dysfunction.
• To assess the algorithm's performance using neuropsychological screening test results.

Main Methods:

• A retrospective study involving 955 participants (341 dementia, 333 MCI, 341 cognitively healthy).
• Utilized data from the Mini-Mental State Examination, Montreal Cognitive Assessment, and Korean Dementia Screening Questionnaire.
• Applied various machine learning algorithms and calculated performance metrics including accuracy, Cohen's kappa, and AUC.

Main Results:

• Machine learning algorithms achieved high screening accuracies: 67.8%–93.5% for MCI, 96.8%–99.9% for dementia, and 75.8%–99.9% for overall cognitive dysfunction.
• Kappa statistics ranged from 0.351 to 1.000, indicating good to excellent agreement.
• The ML models demonstrated statistically superior Area Under the Curve (AUC) values compared to existing screening models.

Conclusions:

• Machine learning algorithms show promise as effective tools for screening cognitive dysfunction.
• These AI-driven approaches can serve as valuable supportive instruments for clinicians in identifying MCI and dementia.
• Further integration of ML in cognitive screening may improve diagnostic efficiency and patient outcomes.