This project aims to investigate the genetic and neurobiological underpinnings of dementia and Alzheimer’s disease by integrating large-scale genetic, neuroimaging, and cognitive data available through the UK Biobank. Despite major advances in identifying genetic risk loci such as APOE and others linked to amyloid processing and neuroinflammation, the mechanisms through which these genetic factors contribute to structural and functional brain changes remain poorly understood.
Our research will address three key questions: (1) Which genetic variants and polygenic profiles are associated with early neuroimaging markers of cognitive decline? (2) Can generative AI models be used to handle missing or incomplete multimodal data and enhance predictive accuracy? (3) How can integrated genetic-neuroimaging models improve differential diagnosis and prognosis in dementia?
The objectives are to identify genetic proxies for neurodegeneration, develop robust generative models to impute missing data, and build interpretable, data-driven tools for predicting cognitive decline and dementia progression. Using genome-wide association analyses (GWAS) of imaging-derived phenotypes (IDPs) and advanced machine learning methods, we aim to uncover the genetic architecture of brain aging and develop predictive models that integrate genomic, imaging, and cognitive data.
The scientific rationale lies in developing a holistic, end-to-end framework for dementia research, linking genes, brain structure, and cognition, to support early detection and targeted intervention. By advancing understanding of genetic influences on brain health, this project will provide valuable insights for public health strategies, risk stratification, and the development of precision approaches to dementia prevention and care.