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: | Omer Noy |
| Institution: | Tel Aviv University |
| Department: | Computer Science |
| Country: | |
| Proposed Analysis: | Estimating individual treatment effects from observational data is essential to precision medicine, assisting practitioners with actionable decisions. Existing causal inference approaches typically estimate treatment effects given observed individuals’ pre-treatment covariates. Recent work (Karlsson et al., 2022) has shown that learning using privileged information (LuPI) can improve prediction in Gaussian-linear dynamical systems. We wish to generalize this idea to study whether and when privileged information can be used to better answer causal questions. In the causal setting, the privileged information is post-treatment time-series samples observed between the time of treatment decision and the future outcome. As opposed to classical supervised learning, in our setting, learners have access to this post-treatment information that is available only during training and unavailable at test time. We present an algorithm for the causal setting and test it on synthetic data. Our goal is to use data extracted from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) database for estimating patient-level treatment effects in Alzheimer’s disease (AD) patients. This could inform treatment recommendations that will lead to overall better outcomes of AD patients. We further wish to investigate whether our proposed approach is preferable to classical learning, leading to better treatment recommendations for AD patients. |
| Additional Investigators |
