Catellani, P., & Piastra, M. (2025). Promoting physical activity in middle-aged and older adults through digital intervention: matching message frame and recipient profile through deep reinforcement learning. Cogent Psychology, 12(1), 2541720.
Although physical activity has a positive impact on physical and mental health, many people remain inactive.
We investigated whether a two-week mobile app intervention would improve the intention to be physically active, especially when messages were tailored to users’ psychological characteristics. A sample of 305 adults aged 50- 80 received prefactual (i.e. ‘If… then…’) messages that differed in terms of valence of gaining or not losing and focus on physical or mental benefits. We measured prior engagement, attitude, intention, regulatory focus and disease susceptibility. Post-intervention, we assessed message evaluation and again intention toward physical activity.
The development of a Dynamic Bayesian Network allowed us to predict message effectiveness as a function of recipient characteristics. Only those who initially had low levels of physical activity and intention improved their intention to be physically active, and only under specific combinations of message framing, prevention focus and disease susceptibility. Finally, by applying Deep Reinforcement Learning to develop an automatic strategy to quickly profile recipients and select the most effective messages.
The integration of psychosocial models and artificial intelligence enables the development of personalized communication strategies to promote physical activity in middle-aged and older people at scale.