Artificial Intelligence for Precision Physical Activity Guidance: Connecting Cohort Evidence With Practice
Med Research · 2026-11-10
DOI: 10.1002/mdr2.70099 ↗Abstract
ABSTRACT Large‐scale population cohorts have expanded the evidence base linking movement behaviours with cardiometabolic disease, cognitive decline, depression, mortality and biological ageing. However, many studies still depend on predefined variables and population‐average models, which limits their value for tailoring guidance to people with different behavioural profiles, health conditions and functional constraints. This correspondence examines how artificial intelligence (AI) could analyse daily movement patterns and combine wearable data with clinical information to inform activity guidance. It also considers how AI could organise and update the evidence used to develop that guidance. Yet, improved predictive performance does not resolve persistent challenges in cohort research, including measurement heterogeneity, limited transportability, residual confounding and the gap between prediction and action. Progress towards intelligent activity prescription will depend on four connected stages: data alignment, model development, decision‐support design and translation into practice. These applications require epidemiological evaluation before they can inform practice.
EUVIMED