Prediction of future cognitive impairment among the community elderly: A machine-learning based approach

Area of Science:

• Gerontology
• Artificial Intelligence
• Public Health

Background:

• Early detection of cognitive impairment (CI) in the elderly is vital for timely intervention.
• Current diagnostic methods (neuroimaging, genetic, cerebrospinal fluid analysis) are costly and invasive.
• Non-invasive prediction using easily accessible variables is needed for community-dwelling older adults.

Purpose of the Study:

• To develop and validate a machine learning-based predictive model for identifying future cognitive impairment.
• To utilize readily available data from community healthcare settings for prediction.
• To enable early screening of cognitive decline in the elderly population.

Main Methods:

• Utilized a nationwide dataset of 3424 community-dwelling elderly individuals initially free of cognitive impairment.
• Employed the Gradient Boosting Machine (GBM) algorithm for predictive modeling.
• Predicted the onset of cognitive impairment within a 2-year timeframe.

Main Results:

• The Gradient Boosting Machine model demonstrated high predictive performance.
• Achieved a sensitivity of 0.967, specificity of 0.825, and Area Under the Curve (AUC) of 0.921.
• Indicated strong potential for machine learning in predicting future cognitive impairment.

Conclusions:

• A machine learning-based model can effectively screen for future cognitive impairment using common healthcare data.
• This approach offers a cost-effective and non-invasive alternative to traditional diagnostic methods.
• Further refinement of the model can significantly improve cognitive function management in community-dwelling elderly.