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: | Chenxi Li |
| Institution: | Michigan State University |
| Department: | Epidemiology and Biostatistics |
| Country: | |
| Proposed Analysis: | Alzheimer’s disease (AD) is one of the most common aging-associated diseases in the world. It is important to estimate the AD’s incidence rate for the timing of AD examinations and disease-modifying interventions and for cost-effective health care planning. Many time-to-event analyses of AD have been conducted to evaluate the prognostic roles of genes, blood and CSF biomarkers as well as demographic, health and lifestyle factors on AD development. With the emerging massive neuroimaging data from Alzheimer’s studies such as ADNI, scientists can predict AD progression more accurately as the data come directly from the affected area of human body. A lot of statistical research has been done in how to use neuro MRI data, both structural (sMRI) and functional (fMRI), to discriminate diseased subjects from normal controls. Less effort has been devoted to predicting future disease occurrence with brain structural or activity patterns. Previous Alzheimer’s research suggested that AD patients have reduced volume of regions of interest such as hippocampus and loss of functional connectivity within some resting-state brain networks compared to healthy controls. This project is going to develop statistical methods for investigating the effects of these sMRI and fMRI features on time to AD and predicting AD incidence with them. The specific aims are as follows. Aim 1: Detect the brain regions of which the tissue density or volume is predictive of time to AD based on sMRI data. We will couple the Cox model with Elastic Net penalty and generalized graph fused LASSO penalty respectively to select concentrated brain voxels whose volumetric value is prognostic of time to AD. Efficient computational algorithms and powerful inferential tools for the regularized ultra-high dimensional Cox models will be developed. Aim 2: Predict time to AD using ICA-based measures of functional connectivity of resting-state networks. Subject-specific measures of intra-network functional connectivity will be obtained by applying multi-subject ICA and dual regression to resting-state fMRI data. Using these intra-network connection strengths as features in survival models on time to AD, regularized by LASSO, SCAD, etc. if high dimensional ICA is employed, we will select the resting-state networks of which the functional connectivity is associated with AD development and predict risk of AD incidence based on their connectivity strengths. The developed approaches will be applied to the data from ANDI for substantive findings. The new knowledge on the association of brain structure and activity patterns with time to AD will facilitate determining the frequency of AD examinations, selecting surrogate endpoints for AD prevention clinical trials, and choosing the timing of disease-modifying interventions. |
| Additional Investigators | |
| Investigator's Name: | Jianrui Zhang |
| Proposed Analysis: | We will develop a post-selection inference method and apply it to ADNI data to test the significance of the predictors of AD selected by variable selection method s. |
