Predicting Dementia Risk for Elderly Community Dwellers in Primary Care Services Using Subgroup-specific Prediction...

Area of Science:

• Gerontology
• Computational Neuroscience
• Public Health

Background:

• Early dementia detection is vital for timely intervention and patient care.
• Predicting dementia risk aids in identifying individuals needing prompt clinical attention.
• Community-based primary care data offers valuable insights for dementia risk assessment.

Purpose of the Study:

• To identify high-risk dementia groups using predictive modeling.
• To predict Mini-Mental State Examination (MMSE) outcomes for dementia risk stratification.
• To enhance prediction accuracy by addressing inter-individual variability.

Main Methods:

• A multi-stage machine learning approach combining supervised and unsupervised methods.
• Data preprocessing included missing value imputation and wrapper-based feature selection (10 of 567 variables selected).
• Optimized hierarchical clustering for data partitioning, followed by subgroup-specific supervised learning models.

Main Results:

• The multi-stage method yielded satisfactory performance in predicting dementia risk classes.
• Subgroup-specific prediction models demonstrated effectiveness in identifying at-risk individuals.
• The approach successfully reduced data complexity while maintaining predictive power.

Conclusions:

• Machine learning models can effectively predict high-risk dementia cases.
• Integrating unsupervised and supervised learning enhances early dementia detection.
• This methodology supports improved clinical decision-making for dementia prevention and management.