There are many active research projects accessing and applying shared ADNI data. Use the search above to find specific research focuses on the active ADNI investigations. This information is requested annually as a requirement for data access.
| Principal Investigator | |
| Principal Investigator's Name: | Yanshan Wang |
| Institution: | Mayo Clinic |
| Department: | Health Sciences Research |
| Country: | |
| Proposed Analysis: | Motivation Alzheimer’s disease (AD) is a degenerative brain disease and the most common form of dementia in United States. AD affects about 5.7 million Americans, and the number will grow to 13.8 million by mid-century. As the number of patients with AD is expected to grow, finding ways to prevent and lower the risk of AD becomes a crucial matter. Most cases of Alzheimer's disease result from a mix of genetic, environmental and lifestyle factors. Recent studies have demonstrated that polygenic risk is predictive of disorder status, and that polygenetic risk scores (PRSs) are associated with symptom profiles within a disorder. Those approaches using PRSs produce promising results, however, have several limitations. First, PRS studies in AD have primarily been validated in “research cohorts”. However, to consider the incorporation of PRSs in clinical care results must first be validated using routine clinical care data. Second, most of these research cohorts are of predominantly European ancestry (EUR), and the application of PRSs to diverse clinical populations requires evaluation and refinement across ancestries. Third, despite the frequent comorbidity of these disorders, most prior studies have evaluated effects of individual-disorder PRSs in trying to optimize prediction. Finally, there has been little research into the combined effects of social determinants of health (SDOH) factors and PRSs on AD, or assessment of SDOH and PRSs on important clinical outcomes such as treatment adherence, behaviors, and healthcare services utilization. To address these critical limitations, the overarching goal of this project is to examine genotype-phenotype relationships using ADNI data. Our objective is to develop improved methods for EHR phenotyping for AD, and evaluate associations between PRSs and these EHR-derived phenotypes as well as psychiatric outcomes and SDOH factors. The project, if successful, will end up with publications and will seek NIH fundings. The Specific Aims are: Specific Aims Specific Aim 1 – Develop digital phenotyping algorithms using EHRs for AD and outcomes. Using ADNI clinical data, we will develop natural language processing (NLP) and machine learning (ML), including deep learning, methods to 1) identify subjects with AD; 2) ascertain treatment adherence and lifestyles; 3) assess quality of life and health economics; and 4) detect cognitive scale changes, e.g., subscale of the Alzheimer’s Disease Assessment Scale, scales that assess AD patients’ abilities to perform ADL (Activities of Daily Living), scales to assess behavioral symptoms in dementia. We will also quantify healthcare utilization in patients with these conditions. Specific Aim 2 – Polygenic prediction of EHR-driven AD’s phenotypes. We will evaluate the performance of PRSs based on results from large research cohorts in predicting the EHR-derived phenotypes for AD using linked EHR-genotype data from the genetics data in ADNI. We will then apply multi-PRS methods to maximize the prediction of the EHR-derived phenotypes and dissect genetic similarities among these comorbid traits and their overlap. We will also examine and maximize the performance of PRSs across ancestries. Finally, we will perform genome-wide association study (GWAS) on the EHR-derived phenotypes and examine the possible pleiotropic effects of known genome-wide significant variants from existing GWASs of these phenotypes. Specific Aim 3 – Evaluation of genetic, images, SDOH predictors of EHR-derived phenotypes and associated clinical outcomes. We will 1) evaluate how PRSs and SDOH factors contribute both independently and interactively to the risk of AD; 2) examine the relationship between PRSs and images using deep learning techniques; 3) explore PRSs and healthcare utilization associations in these disorders, and 4) examine the role of general medical comorbidities in these associations. |
| Additional Investigators |
