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: | Joshua Chuah |
| Institution: | Rensselaer Polytechnic Institute |
| Department: | Biomedical Engineering |
| Country: | |
| Proposed Analysis: | The goal of this analysis is to establish a statistical framework to identify biomarkers from high-dimensional datasets such that there is minimal variability in a diagnostic model using these biomarkers. This is to ensure the biomarkers measured are truly representative of the population when developing a machine learning-based diagnostic test. By measuring the stability of several commonly used feature selection methods such as recursive feature elimination, reliefF, and forward sequential selection, and computing the stability of the features selected by these in response to changes in the training/testing split or in the presence of small but detectable gaussian noise, we can have a better idea of which feature selection methods are better suitable for translation to the clinic. For this, we will need access to biospecimen, specifically tabular data such as proteomics or metabolomics that measure a wide range of potential biomarkers. Additionally, we will also determine how much the model parameters of a machine learning model (i.e., coefficients) change due to noise or training/testing split, and if other factors inherent to data such as missing value imputation strategies also play a role in potential variability in the model. |
| Additional Investigators | |
| Investigator's Name: | Juergen Hahn |
| Proposed Analysis: | Advisor to the aforementioned project |
