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

Area of Science:

• Behavioral science
• Control systems engineering
• Public health

Background:

• Insufficient physical activity (PA) is a significant global health issue.
• PA offers numerous proven health benefits.
• Existing interventions often lack personalization and adaptability.

Purpose of the Study:

• To evaluate Model Predictive Control (MPC) as a framework for personalized adaptive behavioral interventions.
• To enhance physical activity (PA) levels, specifically daily step counts.
• To explore the application of MPC within the Social Cognitive Theory (SCT) framework.

Main Methods:

• Utilized a computational model based on Social Cognitive Theory (SCT) and fluid analogy.
• Developed and proposed diverse control strategies using MPC under various conditions.
• Investigated the impact of measurement limitations, physical/budgetary constraints, and plant limitations on decision-making.

Main Results:

• Demonstrated MPC's potential for delivering feasible and personalized PA interventions.
• Showcased adaptability to conditions with limited measurements, nonlinearity, and plant-model mismatch.
• Highlighted the user-friendly nature of MPC-based interventions.

Conclusions:

• Model Predictive Control (MPC) provides a robust framework for personalized behavioral interventions to increase physical activity (PA).
• MPC effectively addresses complexities such as limited data and system uncertainties.
• This approach holds significant potential for improving public health through enhanced PA levels.