About
We aim to use the data in UK Biobank to explore the relationship and underlying mechanisms between varied types of population characteristics, environmental and behavior exposure, genetic backgrounds and chronic diseases.
Epidemiological studies have linked numerous environmental factors with risk of multiple chronic diseases and death. However, virtually no studies have been done to systematically evaluate the relationships between environmental factors and risk of chronic diseases and death at individual exposure levels with a population of 500,000, not to mention to consider the complex correlation among varied diseases such as multimorbidity and comorbidity patterns. At the same time, with the rapid expand of medical research , intensive, large, multidimensional, and diverse datasets pose many challenges to traditional statistics methods. Machine learning (ML) proved to be a great complement to traditional statistical approaches, providing new perspectives in medical data analysis.
To address these issues, we will use data from UK Biobank to use network analysis to identify and visualize the non-random associations between health conditions, to find disease pattern such as multimorbidity and comorbidity in the population. Conduct prospective observational analyses to exam the association between environmental and genetic factor and health outcomes by using traditional regression analysis and machine learning approaches. We also will explore the casual relationship and potential mechanism between exposures and health outcomes by mediation analyses, phenome-wide association analyses, genome-wide association analyses, mendelian randomization analyses, and metabolomics analyses.