Comparative study of machine learning methods for modeling associations between risk factors and future dementia...

Area of Science:

• Gerontology
• Neuroscience
• Data Science

Background:

• Dementia risk is significantly influenced by modifiable factors, highlighting the importance of identifying key contributors.
• Machine learning (ML) is increasingly utilized in healthcare for predictive modeling, with prior studies exploring its application in dementia progression.
• Understanding cognitive risk factors is crucial for early intervention and prevention strategies.

Purpose of the Study:

• To compare the predictive performance of different machine learning algorithms in modeling the association between known cognitive risk factors and future dementia diagnoses.
• To evaluate logistic regression, random forest, and neural networks for their efficacy in identifying individuals at higher risk of developing dementia.

Main Methods:

• Analysis of a subset from the AGES-Reykjavik Study dataset, including 1,491 older adults initially assessed with healthy cognition.
• Data collection occurred at two time points, approximately five years apart, incorporating demographics, MRI data, and other health information as cognitive risk factors.
• Three machine learning methods—logistic regression, random forest, and neural networks—were employed to model associations with incident dementia cases identified at follow-up.

Main Results:

• The random forest algorithm demonstrated superior performance compared to neural networks and logistic regression in predicting dementia risk based on cognitive factors.
• Performance metrics indicated that ML methods offer enhanced predictive accuracy over traditional statistical approaches for identifying individuals susceptible to dementia.
• The study successfully modeled associations between a range of cognitive risk factors and subsequent dementia development.

Conclusions:

• Machine learning, particularly the random forest algorithm, shows significant potential for accurately predicting dementia risk by analyzing cognitive risk factors.
• These findings suggest that ML-based predictive models can offer more precise identification of individuals at elevated risk for dementia compared to conventional methods.
• Leveraging ML in dementia research can aid in developing targeted prevention and intervention strategies by pinpointing at-risk populations.