Relative importance of multimodal risk factors to discriminate memory trajectories: Similarities and differences across ethnoracial groups in cognitively normal older adults

Background Inequalities in dementia incidence and heterogeneities observed on memory decline are present across ethnic/racial groups. Research suggests that inequities risk/protective factors contribute towards these differences. We examine risk multiple domains to identify those best discriminate differences trajectories ethnoracial Method used an ethnoracially diverse longitudinal cohort of cognitively normal older adults (n = 299; mean age 73.47 (6.90) years; 63.9% female) from the University California, Davis-Alzheimer’s Disease Center. First, we applied latent class growth analysis objectively classify into distinct groups a maximum 10 follow-ups. Second, random forest analyses (RFA) relative importance 22 seven (demographic, genetic, brain morphometry, plasma functional biomarkers, psychological lifestyle factors) low/high performers as stratified by three (African Americans [AA; n 111]; Hispanics [n 108], Whites 66]). Result four classes (low, intermediate, high, very high) represented (Figure 1). for our RFA analyses, collapsed low intermediate group 136; 40.4% AA, 25.6% Hispanics, 28.8% Whites), high performing 148; 34.6% 50% 15.4% Whites). High was predicted 2) AA lower smoking years, Functional Assessment Questionnaire (FAQ), younger age, higher education, neurofilament light (NfL), Apolipoprotein E non-carriers, neuropsychiatric inventory (NPI) severity, men, hippocampal volume (AUC 0.73); with FAQ, Geriatric Depression Scale (GDS), smaller third ventricle, larger volume, NPI lateral NfL 0.86); Whites, alcohol use frequency, NfL, GDS, ventricles, white matter hyperintensities 0.82). Conclusion similarities predictors versus Identifying will promote integration multimodal develop novel complex profiles leading early detection identification asymptomatic or preclinical adults.