Model Predictive Control in mHealth: A Decision Framework for Optimised Personalised Physical Activity Interventions.

Abstract

A major problem in global health is insufficient physical activity (PA) by individuals, despite its proven benefits. In this paper, Model Predictive Control (MPC) is evaluated as the basis for delivering personalised optimal adaptive behavioural interventions aimed at improving PA (in terms of the number of steps walked per day). Utilising the behavioural framework of Social Cognitive Theory (SCT) expressed as a fluid analogy computational model, a series of diverse control strategies are proposed under different circumstances that provide insights into how MPC can serve as a broad-based framework for delivering PA behavioural interventions. The complexities of measurement and information availability, physical and budgetary constraints, and plant limitations and their impact on decision-making are explored, with the results obtained demonstrating MPC's potential to deliver feasible, personalised, and user-friendly behavioural interventions under conditions involving limited measurements, nonlinearity, and plant-model mismatch.