Major chronic diseases including cardiovascular disease, osteoporosis, and metabolic syndrome pose significant public health challenges. Ultrasound examination provides rich tissue information non-invasively. UK Biobank’s extensive ultrasound data enables systematic investigation of ultrasound-derived phenotypes and disease associations. This study addresses: Can ultrasound-derived variables effectively predict chronic disease risk? What is their incremental predictive value beyond traditional risk factors? Can ultrasound phenotypes reveal cross-organ disease mechanisms? Can AI-based feature mining identify novel risk signals?
This study establishes hierarchical research objectives. First, characterize distribution patterns of core parameters including carotid intima-media thickness, plaque scores, heel ultrasound velocity and attenuation coefficient, and their correlations with baseline clinical characteristics. Second, evaluate association strength between ultrasound-derived variables and major chronic disease endpoints including cardiovascular events, metabolic diseases, and skeletal system diseases. Third, construct disease risk prediction models incorporating ultrasound variables and quantify their incremental predictive value compared with baseline models containing only traditional risk factors. Finally, explore interactions between ultrasound phenotypes and traditional clinical indicators, genetic variants, metabolomics and proteomics data to elucidate biological mechanisms of disease occurrence.
Ultrasound demonstrates diverse clinical applications: carotid ultrasound detects early atherosclerosis predicting cardiovascular events, while quantitative heel ultrasound assesses bone quality correlating with fracture risk. UK Biobank’s multi-site ultrasound data enables comprehensive disease association studies. This study integrates ultrasound phenotypes with clinical outcomes to construct risk assessment models for early screening and precision prevention.
