Abstract:Created with BioRender.com. Background A lack of consensus exists in linking demographic, behavioral, and cognitive characteristics to biological stages of dementia, defined by the ATN (amyloid, tau, neurodegeneration) classification incorporating amyloid, tau, and neuronal injury biomarkers. Methods Using a random forest classifier we investigated whether 27 demographic, behavioral, and cognitive characteristics allowed distinction between ATN‐defined groups with the same cognitive profile. This was done separately for three cognitively unimpaired (CU) (112 A‐T‐N‐; 46 A+T+N+/−; 65 A‐T+/‐N+/−) and three mild cognitive impairment (MCI) (128 A‐T‐N‐; 223 A+T+N+/−; 94 A‐T+/‐N+/−) subgroups. Results Classification‐balanced accuracy reached 39% for the CU and 52% for the MCI subgroups. Logical Delayed Recall (explaining 16% of the variance), followed by the Alzheimer's Disease Assessment Scale 13 (14%) and Everyday Cognition Informant (10%), were the most relevant characteristics for classification of the MCI subgroups. Race and ethnicity, marital status, and Everyday Cognition Patient were not relevant (0%). Conclusions The demographic, behavioral, and cognitive measures used in our model were not informative in differentiating ATN‐defined CU profiles. Measures of delayed memory, general cognition, and activities of daily living were the most informative in differentiating ATN‐defined MCI profiles; however, these measures alone were not sufficient to reach high classification performance.

Clinical variables contributing to the identification of biologically defined subgroups within cognitively unimpaired and mild cognitive impairment individuals