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: | Vasiliki Lagou |
| Institution: | University of Surrey |
| Department: | Statistical Multi-Omics, People Centered AI Inst |
| Country: | |
| Proposed Analysis: | Neurodegenerative diseases of the ageing population, such as Alzheimer’s and Parkinson’s diseases are comprised of different subclinical types which are believed to be of distinct poly-omic architecture from each other eg. late onset vs early onset types. For effective intervention studies to happen, targeting the correct patient population, thus, correct subtyping of the disease is of crucial importance. The opportunities for efficient biological subtyping of affected individuals are increasing everyday, first; by the increasing dimensions of available biological datasets and second; by increasing availability of AI methods and their incorporations in omics studies. Our team has earlier identified multiple metabolic and proteomic signatures pointing to various promising peripheral markers CDH6, HAGH, tyrosine, glycylglycine, glutamine, lysophosphatic acid C18:2, and platelet-activating factor C16:0 as well as an obvious effect of APOE genotype on representation of the metabolic networks across case and control groups. We also showed that branched-chain amino acids and HDL particle fractions are associated with change of Alzheimer’s disease risk. In addition to proteomics, genetics and metabolomic studies as exemplified above, gut microbiome research showed evidence supporting Microbiome-Gut-brain axis, a new landscape to be mapped for neurodegenerative studies. We have shown strong dysbiosis for a set of Parkinson’s disease patients and other recent research points out dysbiosis as a hallmark of Alzheimer's disease as well, while the two diseases are now genetically linked. Following up on this evidence we propose to use state-of-art machine learning methods to identify clusters of patients that manifest biological similarities using available public omics data. The overarching research project is composed of two aims: ● Modelling. First aim is to perform an empirical comparison of machine learning approaches on different datasets and use multi-modal methods (supervised and unsupervised) to separate healthy individuals from those affected by neurodegenerative conditions. Moreover, we aim to identify clusters within the diseased group, indicating distinct biological subtypes which are of clinical importance for intervention. While doing so, we will also compare different methods on multi-modal omics data, such comparisons were not previously performed in a systematic way. ● Web-based visualisation. Second aim is to develop a publicly available, web-based tool for performing sample clustering using multi-omic profiles of different modalities. Such web-based services with similar user profiles (e.g http://enterotypes.org, for enterotyping and LD-hub for LD-regression based phenotype correlation analysis) exists for genetic data manipulation/analysis and they are very widely used and appreciated by the community of multi-omics researchers. To this date, a tool for multi-modal omics data based clustering does not exist and we believe such a tool will be empowering the cross-disciplinary interactions between health sciences and AI. |
| Additional Investigators |
