Mueller, T. T.; Starck, S.; Llalloshi, R.; Kaissis, G.; Ziller, A.; Rueckert, D.; Braren, R.

Understanding the process of ageing has become a highly desirable goal in human health and disease research. A multitude of factors impact the way we age and understanding their connection can help with early detection of age-related diseases and give insights into what causes abnormal ageing. In this study, we analyse how medical images can be used as biomarkers for ageing via deep learning techniques. We evaluate ageing in different local body systems (liver, lungs, spine, intestine, muscle, heart), the brain, and across the whole body using 70,000 subjects from the UK Biobank population study. We analyse correlations between lifestyle factors --such as smoking-- or diseases and accelerated ageing. Furthermore, we show differences in survival between subjects with accelerated and decelerated ageing and study the impact of individual body regions on the whole body age by generating a Digital Twin that can aid personalised medicine. Our work gives insights into how medical imaging can be used to identify abnormal ageing both on a local and a global scale and investigates the interplay between different body regions. It can help understand ageing patterns and lay a foundation for future applications of age predictors for risk assessment.