Models for predicting risk of dementia: a systematic review

Area of Science:

• Neuroscience
• Gerontology
• Public Health

Background:

• Dementia risk models are vital for predicting future dementia probability.
• These models can inform targeted interventions to prevent dementia onset.

Purpose of the Study:

• To systematically review and assess the predictive performance of existing dementia risk models.
• To identify common predictors and evaluate the accuracy of dementia risk prediction.

Main Methods:

• Conducted a systematic review of 8462 studies, identifying 61 articles on dementia risk models.
• Assessed model performance using sensitivity, specificity, and area under the curve (AUC) from receiver operating characteristic analysis.

Main Results:

• Most models focused on late-life risk (n=39), followed by mild cognitive impairment to Alzheimer's disease (n=15).
• Common predictors include age, sex, education, cognitive tests, BMI, alcohol, and genetics.
• Models generally showed moderate-to-high predictive ability (AUC > 0.70), with the highest AUC of 0.932 for mild cognitive impairment prediction.

Conclusions:

• Existing dementia risk models demonstrate acceptable predictive accuracy.
• The development of population-specific dementia risk models is essential for diverse populations and subpopulations.