Machine Learning on Densitometric Imaging for Novel Cardiometabolic & Musculoskeletal Health Phenotypes of Disease and their Developmental Origins
Bone densitometry scans via dual-energy x-ray absorptiometry is already implemented in the community as a diagnostic tool for fracture risk assessment. Yet, these machines also provide images with underexplored potentially clinically important information regarding individual's cardiovascular and musculoskeletal health that are currently underexplored and underutilized. The aim of this study is therefore to use machine learning to develop and test algorithms to provide new ways to identify who will go onto developing cardiovascular disease, falls, fractures and dementias. Furthermore we seek to determine the genetic and environmental factors related to the development of these phenotypes to prevent the development and progression of these phenotypes before the onset of clinical events. These algorithms may also help clinicians identify individuals who are most at risk for functional decline, falls, fractures and cardiovascular diease enabling early referral to preventative and treatment strategies. Data from this project has the capacity to redefine the identification and subsequent management of musculoskeletal and cardiovascular health. The wealth of data from the UK Biobank will serve as an ideal resource to examine this research question that we will aim to address over a 36-month period.