A retrospective multi-centre radiomics study with internal and external validation showing moderate discriminatory performance (AUC 0.790–0.818 on external cohorts) for a clinically relevant surrogate endpoint, but lacking prospective design, hard clinical outcomes, or demonstration of impact on treatment decisions.
A sound cross-sectional study in a large, well-characterized HBsAg-positive cohort with rigorous discrimination analysis, but cross-sectional design and surrogate endpoint (ultrasound steatosis) limit the strength; findings require prospective validation before clinical adoption.
Hospital-based cross-sectional study with large sample and multivariate analysis showing age-specific associations between metabolic phenotypes and angiographically confirmed CAD, but lacks a prospective design and causal inference; findings in young adults are novel but require validation.
Early-stage mechanistic study combining public database analysis, cell line knockdown, and mouse models with surrogate endpoints (proliferation, migration, bone metastasis incidence) but lacking human clinical validation of the proposed pathway.