Application of latent and integrative approaches to understand the spectrum of cognitive aging

点击翻译按钮获取中文摘要

ABSTRACT The proposed research project and my mentoring and career development plan will enable me to become a successful independent investigator in cognitive aging and cognitive epidemiology with a focus on longitudinal statistical modeling of cognitive function, harmonizing datasets using coordinated integrative analytic approaches, epidemiology, and intensive cognitive assessment. Non normative cognitive decline is a major issue in our aging society; amongst the many dementias, Alzheimer’s disease (AD) alone is estimated to affect 17% of individuals between the ages of 75 and 84, and accounts for $226 billion for our annual health-care budget without accounting for other dementias, some of which are potentially reversible, with mixed etiology. The heterogeneity among the presymptomatic stages of Alzheimer’s disease and related dementias contributes to the limited effectiveness of clinical treatment and prevention trials. To this end, the proposed research project and training plan will investigate individual differences in cognitive trajectories in older adults; it will also characterize individual differences in vascular risk factors, physical function, and emotional states, using techniques from structural equation modeling, latent variable modeling, and Bayesian approaches. The training plan includes didactic course work, experiential learning, and structured tutorials with well-recognized mentors in various areas of their expertise. The research plan takes advantage of a large and rich NIH-funded longitudinal dataset, the Einstein Aging Study (EAS) to parse heterogeneity in cognitive decline across individuals and to identify factors that are associated with different trajectories. Through the use of latent trajectory models fit to data from EAS participants to partition heterogeneity and to identify candidates who are eligible for early intervention we will be able to establish the type of intervention that may benefit specific profiles. We propose to use prediction models that employ vascular risk factors (including diabetes, CVD, and hypertension), physical function (gait speed, grip strength, lung function, and frailty), and measures on emotional states (stress, depression, and cortisol) because these can be controlled or reversed, and because they are also associated with cognitive impairment. By classifying individuals, we will be better able to predict who will benefit from specific treatments. By harmonizing our results and replicating our findings in other datasets that are part the NIH-funded network, the Integrative Analysis of Longitudinal Studies on Aging (IALSA), our results will comply with the new NIH guidelines on Rigor and Reproducibility; our results will also have increased generalizability. Our project will thus have implications for both behavioral and pharmacological interventions. These outcomes will impact mild cognitive impairment (MCI), presymptomatic Alzheimer’s disease, and dementia research, especially in furthering forward classification and early detection, and in designing targeted intervention programs for individuals at high risk. Key words: cognitive epidemiology; individual differences; heterogeneity; MCI; Alzheimer’s disease; vascular dementia; longitudinal; integrative analysis; coordinated analysis