Multimorbidity

Although clinical guidelines have traditionally focused on single chronic conditions, most adults in later life live with multimorbidity, the co-occurrence of multiple long-term conditions (MLTCs). By leveraging longitudinal primary care data from millions of individuals in the UK, we develop deep learning-based clustering methods and interpretable machine learning methods to move beyond simple disease counts and provide a comprehensive understanding of health across the life course. Our research has identified distinct multimorbidity profiles and unique progression pathways while uncovering the social and biological determinants that drive healthcare burden. Central to our work is the discovery of a universal scaling law, a two-parameter power-law relationship where the scaling coefficient and exponent determine the rate and acceleration of disease accumulation. By quantifying how these parameters vary by index condition and socioeconomic deprivation, we identify hidden subtypes and snowballing trajectories within multimorbid populations. Ultimately, these AI-based frameworks aim to transform our understanding of disease onset and progression, enabling more holistic, integrated care and targeted public health interventions.