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: | sharvari gujja |
| Institution: | Wuxi NextCODE |
| Department: | Advanced Artificial Intelligence Research Lab |
| Country: | |
| Proposed Analysis: | We have spent the first year of access to ADNI data working on two major projects which are in progress. For the first project, data analysis includes application of LASSO and deep learning methods (DCNN) on PET data to classify AD patients from controls. We have successfully distinguished these two groups at very high accuracy, as others have done, and we have been exploring which of the imaging and anatomical features are most important drivers of classification. For the second project, we are leveraging molecular data types - whole genome variant data, RNA expression data, DNA methylation for the application of LASSO and deep learning methods (DANN and DBNN) to classify MCI/AD patients from controls and MCI from AD patients. We have applied novel strategies for reducing the complexity of integrated data types using either apriori biological knowledge or dataset co-correlation. We have also used bioinformatics tools for data preprocessing such as normalization and to check for batch effect for genetic data. We want to further train and test these methods to define the most important gene or pathway drivers of the classifications |
| Additional Investigators | |
| Investigator's Name: | Tom Chittenden |
| Proposed Analysis: | We have spent the first year of access to ADNI data working on two major projects which are in progress. For the first project, data analysis includes application of LASSO and deep learning methods (DCNN) on PET data to classify AD patients from controls. We have successfully distinguished these two groups at very high accuracy, as others have done, and we have been exploring which of the imaging and anatomical features are most important drivers of classification. For the second project, we are leveraging molecular data types - whole genome variant data, RNA expression data, DNA methylation for the application of LASSO and deep learning methods (DANN and DBNN) to classify MCI/AD patients from controls and MCI from AD patients. We have applied novel strategies for reducing the complexity of integrated data types using either apriori biological knowledge or dataset co-correlation. We have also used bioinformatics tools for data preprocessing such as normalization and to check for batch effect for genetic data. We want to further train and test these methods to define the most important gene or pathway drivers of the classifications |
