Musculoskeletal disorders (MSD) arise from complex interactions between ageing, genetics, environment and metabolism. Large resources like UK Biobank & its multimodal dataset-including longitudinal health records, imaging-derived phenotypes, genomics, proteomics and detailed metabolite and clinical measurements, enables high-resolution investigation of these traits.
The transition from physiological aging-related decline to early pathology is poorly understood. Distinguishing these states requires integrating longitudinal phenotypes (e.g., bone mineral density, muscle composition) with imaging-derived structural traits and clinical measurements. Advances in machine learning (ML) and multi-omics now allow for the identification of latent disease subtypes and molecular pathways driving musculoskeletal deterioration and related systemic conditions, which may help us reveal early molecular determinants, disease endo-phenotypes, and biological pathways driving musculoskeletal deterioration and related systemic conditions.
Research Questions
1. Do multi-omics profiles improve disease risk prediction compared to traditional clinical risk factors and how to integrate imaging, clinical and omics (genomics, proteomics, metabolomics) to understand disease progression?
2. How do longitudinal musculoskeletal traits change with ageing versus early pathological decline?
3. Can multimodal health data integration identify early indicators of MSD onset or predict disease risk?
Objectives
1. Characterise trajectories of musculoskeletal traits using longitudinal and imaging data to distinguish age-related changes (e.g. physical measure, accelerometry, BMD, muscle composition) from early pathological decline that may signal disease initiation independent of chronological ageing.
2. Harmonise EHR, imaging, and omics data to capture comprehensive disease phenotypes.
3. Apply ML-based multiomics integration to identify disease biomarkers & progression patterns
