Advancing women’s care in Alzheimer’s disease and other dementias through EHR - The aging population has led to an increase in cognitive impairment (CI) such as mild cognitive impairment or dementia. As the aging population grows, the need for effective diagnostic, management, and predictive tools is more urgent. With this landscape, there is great need for precision medicine approaches considering individual factors, such as sex, in the understanding and management of dementia. In particular, women have a disproportionate burden of dementia. However, the examination of women-specific differences has not been well integrated into precision medicine approaches, despite the extensive literature demonstrating sex differences in brain structure and function over the lifespan. Women-specific factors earlier in life, such as menopause, menopausal hormone therapy, pregnancy, and hypertensive disorders during pregnancy, are hypothesized to affect cognitive health and dementia. Some studies have identified a stronger association between biomarkers, such as apolipoprotein E (APOE) ε4 genotype, and dementia or cognitive decline in women compared to men. Additionally, the progression of dementia in women can also be uniquely influenced by a myriad of biological, social, and environmental variables. Despite the high prevalence of dementia in women, disparities in diagnosis, access to healthcare, clinical trials, treatment, and care management have been reported. Electronic health records (EHRs), particularly clinical free text, contain invaluable information about patients’ medical history to track longitudinal health conditions. Yet, EHRs have not been fully utilized to systematically capture or distinguish women-specific health events, trends, or patterns that could be invaluable in understanding CI progression. To address this gap, we will analyze women-specific health patterns from EHRs that can address diagnostic disparities in CI progression. We will study these questions using: (a) the population-based Mayo Clinic Study of Aging cohort (n=6,531) with longitudinal cognitive assessments and extensive clinical characterization, providing an ideal gold standard for CI; and (b) the Rochester Epidemiology Project, which links longitudinal EHRs for a large Midwestern population (~1.4 million people in 27 counties). Additionally, we will use nationwide data from National Alzheimer’s Coordination Center to increase representation of the population and to enhance the generalizability. The primary goal of this project is to develop an informatics tool to extract women’s health conditions from multi-site EHRs (Aim 1), analyze differences in CI symptoms, diagnosis, and progression in women in relation to biomarkers (Aim 2), and develop fair and explainable predictive models for CI, assessing the prognostic value of women’s health indicators to guide targeted health adjustments (Aim 3). The proposed tools and methodologies will address diagnostic and treatment disparities in women, ensuring transparent and fair healthcare decisions. The proposed aims align closely with the imperative need and could significantly impact the lives of women at risk for Alzheimer’s disease and other dementias.
