Building intelligent activity prescription: a roadmap across data, modelling and translation (IMAGE)
Caption
The proposed pathway comprises four linked stages. First, data alignment involves harmonising cohort measures, standardising device-processing procedures, aligning time scales and outcomes and incorporating richer inputs across physical activity, sedentary behaviour, sleep and related health domains. Second, model development combines 24-hour behavioural representation, multimodal integration, external validation, interpretability, updating and knowledge extraction under expert-defined rules. Third, decision-support design integrates behavioural patterns, sleep background, chronic disease status, functional capacity and practical constraints to prioritise actionable options, such as increasing movement, strength training, improving sleep or interrupting prolonged sitting. Fourth, translation into practice requires understandable recommendations, feasible delivery, clinical or public-health integration and continuous evidence updating.
Credit
Yanwei You. Reproduced from You Y. Med Research. Published online November 10, 2026. doi:10.1002/mdr2.70099. © 2026 The Author(s). Creative Commons Attribution License (CC BY).
Usage Restrictions
none
License
Original content