Scientific Rationale
Cardiovascular, renal, metabolic, and cognitive (CRMC) diseases and related mortality pose gobal health challenges. Current risk assessment mainly relies on single-dimensional clinical indicators or limited omics, insufficient to capture the complex mechanisms of diseases or guide precise intervention. Gaps remain in identifying novel biomarkers and defining new clinical phenotypes in the CRMC field, hindering understanding of disease heterogeneity. Integrating multi-dimensional data-including clinical phenomics, imaging, exposomics, genomics, proteomics, metabolomics, and established biomarkers-is essential to advance CRMC risk assessment and intervention.
Objectives & Research Questions
This study aims to:
1. Examine associations of novel phenomic markers, lifestyle, biomarkers, imaging, and genetic factors with CRMC outcomes and mortality.
2. Elucidate causal pathways linking these determinants to adverse outcomes.
3. Develop and validate machine learning based predictive models integrating multi-dimensional data.
4. Translate findings into a prototype web tool to support research translation and inform clinical and public health practice.
Methods
We will analyze UK Biobank participants with linked primary care records, physical activity and sleep measures, imaging, genetics, and long-term outcomes. Novel phenomic metrics, their trajectories and cumulative burden, will be assessed using advanced statistical learning for CRMC risk stratification. Causality will be examined via Mendelian randomization and causal machine learning. Deep learning algorithms and AI models will support risk prediction and bioage-phenotype modeling. Interpretability analyses will identify key features, interactions, and co-existing patterns. Internal validation and, where possible, external validation in independent cohorts will be conducted. Prototype web tools will be developed for demonstration and clinical translation, independent of raw UK biobank data.
