Stratifying dementia risk factors: A prediction model and hypothesis-driven analysis.

Area of Science:

• Gerontology
• Neuroscience
• Epidemiology

Background:

• Older adults frequently present with multimorbidity, complicating dementia risk assessment.
• Individual analysis of dementia risk factors may not reflect real-world complexity.
• Methodological comparisons are needed to understand simultaneous risk factor effects.

Purpose of the Study:

• To compare hypothesis-driven and data-driven approaches for dementia risk stratification.
• To identify key dementia risk factors using simultaneous multifactorial assessment.
• To evaluate dementia risk in a real-world cohort.

Main Methods:

• Analysis of 9606 participants from the National Alzheimer's Coordinating Center (NACC) Uniform Data Set (2005-2023).
• Application of machine learning with interpretability analysis and survival models.
• Simultaneous evaluation of 13 potential dementia risk factors.

Main Results:

• 877 participants (9%) developed dementia during a mean follow-up of 6 years.
• Both analytical approaches consistently identified age, depression, and low education as key dementia predictors.
• Higher body mass index was found to be unexpectedly protective against dementia conversion.

Conclusions:

• Age, depression, and low education are robust dementia risk factors, irrespective of analytical methodology.
• Convergent findings support the simultaneous assessment of multiple risk factors for dementia.
• Understanding complex interactions among dementia risk factors is crucial for clinical practice.