Principal Investigator: Dr Jiook Cha
New York State Psychiatric Institute, New York, USATags: 32575, Alzheimer's Disease, connectome, Dementia, Machine Learning, phenome-wide association
Our overarching goal is to develop data-driven Machine Learning-based predictive modeling for AD. Firstly, we aim, using genetic and health records data, to determine risk and protective factors for neurodegenerative diseases (e.g., AD). Secondly, we aim to develop deep learning algorithms for brain imaging markers that are associated with the suggested factors and predict risk for AD. Thirdly, we will test the predictive validity and generalizability of the initial deep learning model by retraining it for an independent clinical data. In sum, this study will generate novel machine learning methodologies for brain imaging analysis informed by multi-dimensional data. This research meets UK Biobank’s purpose because the target disease Alzheimer’s disease and dementia is a significant societal problem and of public interest. Particularly, we aim to develop novel data analytics based on state-of-the-art Artificial Intelligence (e.g., deep convolutional neural network) that has yet to be widely applied to brain imaging data. This development will impact not only Alzheimer’s Disease research but also research of other brain diseases or psychiatric disorders. Data used in Aim #1 includes genetic, electronic health/medical records, emotion/cognition/psychological, and other health-related data (e.g., physical activity, carotid ultrasound, and vision).
Please note that we amended the original application to include the full cohort of participants, without the age exclusion criterion. We are thus requesting data from individuals who have either brain MRI data, genetics, emotion/cognition/psychological data, or miscellaneous data to be used in deep phenotyping for Alzhimer’s disease related dementia as well as other types of dementia.