From self-quantification to predictive ageing: digital twins as a framework for healthy ageing and longevity

Recent advances in biomarker profiling, wearable sensing, and high-density longitudinal self-monitoring have generated detailed individual-level datasets demonstrating that many ageing-associated measures, including cardiometabolic markers, inflammatory signals, functional performance metrics, and epigenetic age estimates remain highly modifiable in adulthood. However, these data also reveal substantial heterogeneity in response timescales and durability, with rapid reversibility of many biomarkers occurring alongside relative stability of structural and functional traits. In the absence of formal temporal and mechanistic models, such observations remain difficult to interpret and are frequently conflated with modification of the underlying ageing process itself. Here, we propose a formal digital twin framework for healthy ageing and longevity that treats ageing as a latent dynamical system rather than a collection of static biomarkers. Through critical synthesis of longitudinal ageing and digital-twin evidence, we derive a testable state-space model that distinguishes observable physiological change from latent ageing dynamics. This framework enables simulation of ageing trajectories, counterfactual testing of interventions, and explicit separation of transient biomarker optimisation from durable changes in ageing dynamics. We describe the model architecture, key state variables, system coupling across biological domains, and strategies for longitudinal validation. By grounding digital twin design in the structure of real high-frequency ageing data, this approach provides a predictive and mechanistically interpretable foundation for modelling health span and longevity.