Related Experiment Video
Updated: May 4, 2026
A Machine Learning Approach to Design an Efficient Selective Screening of Mild Cognitive Impairment
Published on: January 11, 2020
Methodological adaptations for applying the Harmonised Cognitive Assessment Protocol diagnostic algorithm in the
Nicola Ann Ward1, Calum Marr1, Stanley Simoes1
1Centre for Public Health, Queen's University Belfast, Belfast, UK.
Objective:
Nationally representative data are essential for understanding the causes, consequences, and costs of dementia and mild cognitive impairment (MCI) and for informing policy and care planning. This study aimed to describe methodological considerations in applying the Health and Retirement Study Harmonised Cognitive Assessment Protocol (HRS-HCAP) cognitive domain structure and diagnostic algorithm to the Northern Ireland Cohort for the Longitudinal Study of Ageing (NICOLA-HCAP), and to generate harmonised estimates of the prevalence of dementia and mild cognitive impairment (MCI) in a nationally representative sample.
Results:
A total of 1,037 participants aged ≥ 65 years completed NICOLA-HCAP. Five cognitive domains were identified, all loading onto a second-order general cognitive performance factor: orientation (0.903), memory (0.855), executive function (0.893), language-fluency (0.962), and visuospatial ability (0.812). Model fit was acceptable (SRMR = 0.065; RMSEA = 0.047; CFI = 0.916; TLI = 0.906). Following classification, 6.2% of participants were classified with dementia and 15.8% as MCI. Methodological modifications addressed software differences, normative sample derivation, and cohort-specific adjustments. These findings provide preliminary support for HCAP as a framework for producing harmonised estimates of cognitive status. NICOLA-HCAP will facilitate future investigation of modifiable risk factors for dementia in community-dwelling older adults. Validation studies are required to determine whether resulting classifications are fit for purpose.
