Osteoporotic fractures are a leading cause of premature death in older adults with or without osteoporosis; yet, factors driving excess mortality beyond the fracture itself remain poorly understood. We hypothesise that modifiable lifestyle exposures-especially overall dietary pattern-alter long-term survival through their effects on bone micro-architecture, circulating metabolites and inflammatory proteins.
Research questions
Q1 Which dietary patterns and physical activity profiles are prospectively associated with all-cause and cause-specific mortality among participants with diagnosed osteoporosis?
Q2 Do metabolic and proteomic signatures mediate or moderate these associations?
Q3 Can a multimodal machine-learning model combining lifestyle, biomarker and genetic data identify individuals at highest mortality risk and quantify the benefit of realistic lifestyle changes?
Objectives
Derive data-driven dietary patterns (using repeated 24-h recall and FFQ) and other lifestyle factors.
Estimate hazard ratios for mortality across lifestyle strata in individuals with incident or prevalent osteoporosis (ICD-10 codes, self-report, DXA).
Perform metabolomic and proteomic mediation analyses to pinpoint key molecular pathways.
Train and internally validate gradient-boosting and neural-network survival models that integrate lifestyle, multi-omics and polygenic risk scores.
Translate model outputs into population-attributable fractions and simulate the mortality impact of achievable lifestyle shifts.
Scientific rationale
UK Biobank is uniquely suited because it offers prospectively ascertained lifestyle data, serial imaging, multi-omics assays and virtually complete mortality linkage in the same large cohort. Clarifying the lifestyle-omics-mortality axis will (i) shift osteoporosis prevention from fracture-focused to survival-focused strategies, (ii) generate evidence for precision lifestyle interventions, and (iii) reveal novel biomarkers for therapeutic targeting.
