| Title: | Development of a midlife-specific CogDrisk algorithm (CogDrisk-ML) to enable validated implementation of dementia risk assessment from midlife to late life |
| Journal: | Age and Ageing |
| Published: | 1 Jul 2025 |
| Pubmed: | https://pubmed.ncbi.nlm.nih.gov/40685637/ |
| DOI: | https://doi.org/10.1093/ageing/afaf201 |
Publication 16358
WARNING: the interactive features of this website use CSS3, which your browser does not support. To use the full features of this website, please update your browser.
Abstract
BACKGROUND: Existing dementia risk assessment tools, such as The Australian National University Alzheimer's Disease Risk Index (ANU-ADRI), LIfestyle for BRAin health (LIBRA) and Cognitive health and Dementia Risk Assessment (CogDrisk), show limited validation for middle-aged adults (age 40-64 years). The Cardiovascular Risk Factors, Aging, and Incidence of Dementia (CAIDE) tool, developed almost two decades ago, demonstrated moderate predictive accuracy. As key modifiable dementia risk factors emerge in midlife, there is a need for a new, more accurate midlife dementia risk assessment tool.</p>
OBJECTIVES: To develop CogDrisk-ML, a midlife dementia risk assessment tool that can complement the existing CogDrisk tool for late-life dementia risk assessment.</p>
DESIGN AND SETTINGS: Data from the UK Biobank and the Atherosclerosis Risk in Communities (ARIC) study were used to develop and validate CogDrisk-ML, which was also externally validated using the Whitehall II cohort.</p>
PARTICIPANTS AND COVARIATES: Participants without dementia at baseline were included, with CogDrisk predictors along with additional midlife risk factors based on recent evidence.</p>
MAIN OUTCOME MEASURES: Cox regression models estimated the relationship between risk factors and dementia for each sex. A random-effects meta-analysis model aggregated cohort- and sex-specific regression coefficients to develop CogDrisk-ML. Harrell's C statistics measured predictability, with multiple imputation used for missing data.</p>
RESULTS: CogDrisk-ML outperformed CAIDE in the UK Biobank and Whitehall II cohorts; however, it provided similar C statistics in the ARIC dataset. C statistics (95% confidence interval) for CogDrisk-ML were 0.71 (0.69, 0.74) for the ARIC, 0.75 (0.73, 0.77) for the UK Biobank and 0.70 (0.62, 0.79) for the Whitehall II study.</p>
CONCLUSION: The novel CogDrisk-ML for assessing dementia risk in midlife offers improved predictive accuracy. Combined with the CogDrisk tool for late life, it provides a comprehensive framework for dementia prevention throughout the life course.</p>
15 Keywords
- Adult
- Age Factors
- Aged
- Algorithms
- Decision Support Techniques
- Dementia
- Female
- Humans
- Male
- Middle Aged
- Predictive Value of Tests
- Reproducibility of Results
- Risk Assessment
- Risk Factors
- United Kingdom
5 Authors
- Hamidul Huque
- Heidi Jane Welberry
- Ranmalee Eramudugolla
- Nicola T Lautenschlager
- Kaarin J Anstey
1 Application
| Application ID | Title |
|---|---|
| 79620 | Validation of two dementia risk scores: the revised ANU ADRI and a new risk assessment tool (CogDrisk). |
Enabling scientific discoveries that improve human health