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: | Jonathan Herrera |
| Institution: | Massachusetts Institute of Technology |
| Department: | MIT Computer Science & Artificial Intelligence Lab |
| Country: | |
| Proposed Analysis: | We are a team of undergraduate and graduate students at MIT developing a diagnostic model for AD using data collectable from living subjects (genotype, blood and CSF) as an alternative method to diagnosing from cognitive impairment for pre-mortem AD diagnosis. An accurate way to diagnose AD before death is important for improving medical care, prognosis, and prophylactic care, including tailoring accurate drug prescriptions, providing lifestyle change advice, and running effective clinical trials. There is still no way to firmly diagnose Alzheimer’s Disease (AD) until post mortem brain autopsy. There exist many different types of dementia, and the drug for one type of dementia usually has no effect on other types of dementia due to biological differences. Current clinical methods for identifying what type of dementia a living patient has often cannot distinguish between AD and other types of dementia, in particular in early stages of the disease. Drug development research and clinical studies will benefit greatly from more accurate classification of dementia subtypes from non-invasive data. In order to conduct this analysis we would be using clinical, biospecimen, and genetic data from the ADNI dataset. We will extract genotyping data, blood gene expression data and CSF epigenomic data and merge them with necessary clinical data to form a sub dataset. We will separate the dataset into training, test, and validation sets to facilitate the verification of the models. We will train machine learning models, including SVMs and decision trees, to classify Alzheimer’s state in patients from the ADNI dataset. We will utilize a multiclass classifier to diagnose whether patients have Cognitively Normal (CN), Mild Cognitive Impairment (MCI), and Alzheimer’s Disease (AD). Finally, we will analyze the effectiveness of each method, using metrics such as average classification accuracy (ACC), average area under curve (AUC), and rates of false positives and false negatives. Using this data, we will verify whether this approach would be reliable and useful for clinical studies and other applications. This analysis is also useful in determining whether multi-omics play a significant role in the pathology of AD, and which factors are most effective predictors for classification of dementia. |
| Additional Investigators | |
| Investigator's Name: | Xueying Huang |
| Proposed Analysis: | We are a team of undergraduate and graduate students at MIT developing a diagnostic model for AD using data collectable from living subjects (genotype, blood and CSF) as an alternative method to diagnosing from cognitive impairment for pre-mortem AD diagnosis. An accurate way to diagnose AD before death is important for improving medical care, prognosis, and prophylactic care, including tailoring accurate drug prescriptions, providing lifestyle change advice, and running effective clinical trials. There is still no way to firmly diagnose Alzheimer’s Disease (AD) until post mortem brain autopsy. There exist many different types of dementia, and the drug for one type of dementia usually has no effect on other types of dementia due to biological differences. Current clinical methods for identifying what type of dementia a living patient has often cannot distinguish between AD and other types of dementia, in particular in early stages of the disease. Drug development research and clinical studies will benefit greatly from more accurate classification of dementia subtypes from non-invasive data. In order to conduct this analysis we would be using clinical, biospecimen, and genetic data from the ADNI dataset. We will extract genotyping data, blood gene expression data and CSF epigenomic data and merge them with necessary clinical data to form a sub dataset. We will separate the dataset into training, test, and validation sets to facilitate the verification of the models. We will train machine learning models, including SVMs and decision trees, to classify Alzheimer’s state in patients from the ADNI dataset. We will utilize a multiclass classifier to diagnose whether patients have Cognitively Normal (CN), Mild Cognitive Impairment (MCI), and Alzheimer’s Disease (AD). Finally, we will analyze the effectiveness of each method, using metrics such as average classification accuracy (ACC), average area under curve (AUC), and rates of false positives and false negatives. Using this data, we will verify whether this approach would be reliable and useful for clinical studies and other applications. This analysis is also useful in determining whether multi-omics play a significant role in the pathology of AD, and which factors are most effective predictors for classification of dementia. |
| Investigator's Name: | Lilly Edwards |
| Proposed Analysis: | We are a team of undergraduate and graduate students at MIT developing a diagnostic model for AD using data collectable from living subjects (genotype, blood and CSF) as an alternative method to diagnosing from cognitive impairment for pre-mortem AD diagnosis. An accurate way to diagnose AD before death is important for improving medical care, prognosis, and prophylactic care, including tailoring accurate drug prescriptions, providing lifestyle change advice, and running effective clinical trials. There is still no way to firmly diagnose Alzheimer’s Disease (AD) until post mortem brain autopsy. There exist many different types of dementia, and the drug for one type of dementia usually has no effect on other types of dementia due to biological differences. Current clinical methods for identifying what type of dementia a living patient has often cannot distinguish between AD and other types of dementia, in particular in early stages of the disease. Drug development research and clinical studies will benefit greatly from more accurate classification of dementia subtypes from non-invasive data. In order to conduct this analysis we would be using clinical, biospecimen, and genetic data from the ADNI dataset. We will extract genotyping data, blood gene expression data and CSF epigenomic data and merge them with necessary clinical data to form a sub dataset. We will separate the dataset into training, test, and validation sets to facilitate the verification of the models. We will train machine learning models, including SVMs and decision trees, to classify Alzheimer’s state in patients from the ADNI dataset. We will utilize a multiclass classifier to diagnose whether patients have Cognitively Normal (CN), Mild Cognitive Impairment (MCI), and Alzheimer’s Disease (AD). Finally, we will analyze the effectiveness of each method, using metrics such as average classification accuracy (ACC), average area under curve (AUC), and rates of false positives and false negatives. Using this data, we will verify whether this approach would be reliable and useful for clinical studies and other applications. This analysis is also useful in determining whether multi-omics play a significant role in the pathology of AD, and which factors are most effective predictors for classification of dementia. |
