Identification of Dementia & Mild Cognitive Impairment in Chinese Elderly Using Machine Learning

Area of Science:

• Gerontology
• Neurology
• Artificial Intelligence

Background:

• Dementia and mild cognitive impairment (MCI) pose significant global health challenges.
• Early and accurate identification is crucial for timely intervention and management.

Purpose of the Study:

• To evaluate the efficacy of Machine Learning (ML) in identifying critical factors for dementia and MCI.
• To pinpoint key indicators that contribute to the accurate diagnosis of these cognitive conditions.

Main Methods:

• Utilized data from 371 elderly individuals, including demographic and 35 features from 10 assessment scales.
• Employed five ML classifiers with feature extraction, selection, model training, and performance assessment.
• Applied Information Gain and Meta-analysis for feature refinement.

Main Results:

• The Random Forest model demonstrated exceptional performance with an Area Under the Curve (AUC) of 0.961 and accuracy of 0.894.
• Identified three key training features and four meta-features crucial for accurate prediction.
• Highlighted the model's high accuracy in distinguishing between dementia and MCI.

Conclusions:

• Machine learning is a valuable tool for the identification of dementia and mild cognitive impairment.
• Clinical Dementia Rating (CDR) and Neuropsychiatric Inventory (NPI) scale data were identified as critical for Random Forest model training.
• Information Gain and Meta-feature analysis are effective in determining indicative factors for cognitive impairment.