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: | Li-Chu Chien |
| Institution: | Kaohsiung Medical University |
| Department: | Center for Fundamental Science |
| Country: | |
| Proposed Analysis: | We will apply for the Alzheimer’s Disease Neuroimaging Initiative (ADNI) data to examine the usefulness of our proposed method on model selection techniques for the analysis of ordinal longitudinal data with missing values. A short paragraph describing our research is given below. In many longitudinal clinical studies, measurements on the same individual are recorded repeatedly over time. These measurements with the same individual are thus correlated. Ordinal outcomes are typically experienced in these longitudinal clinical studies. In the procedure of recording, a proportion of the measurements may be missed in one or several follow-up assessments. Hence, missing data often occur in longitudinal clinical studies, which potentially lead to invalid inference when no correction is considered to be made. However, to our knowledge, there are still no studies in the literature that discusses the model selection criterion for the missing longitudinal ordinal data. This in turn motivates us to consider how to apply the model selection techniques to analyzing the missing ordinal outcomes in the longitudinal clinical data. Precisely, our proposal attempts to propose a semi-parametric model selection criterion to fill this gap. In this research, we will propose a model selection criterion for incomplete longitudinal ordinal data with ignorable dropouts. The proposed missing longitudinal ordinal information criterion (MLOIC) is based on the generalized estimating equations (GEE) under the proportional odds assumption for selection of the mean model and for selection of the correlation structure model for the longitudinal ordinal outcomes, when the outcome data are observed under monotone dropouts pattern and non-monotone missingness. The proposed MLOIC, based on the expected quadratic loss and the weighted GEE estimation, simultaneously considers a measure of goodness-of-fit of the model and an additional penalization for overfitting the model. Therefore, the proposed MLOIC will be expected to be effective for variable selection in the mean model and selecting the correlation structure for the missing longitudinal ordinal outcomes. Here we would like to apply for the user account to access to the ADNI data. We will use a longitudinal clinical study proved from the ADNI data to examine the usefulness of our proposed model selection method on the incomplete longitudinal ordinal data with ignorable dropouts. |
| Additional Investigators |
