Psychological and technological predictors of physical activity intention-behavior gap: an explainable machine learning analysis

奕蓉 李建光 蔡^*

• School of Physical Education, Hunan University of Science and Technology, Xiangtan, China

Objective: Intention is widely regarded as the most proximal predictor of behaviour. But, physical activity (PA) intentions do not invariably translate into actual exercise behaviour, leaving a intention-behaviour (I-B) gap. The study integrates psychological and technological frameworks to examine the mechanisms that moderate the PA I-B gap. Methods: Unlike traditional dichotomous measures of the PA I-B gap, this study employs baseline correction to derive a standardized continuous measure that quantifies the magnitude of the gap. Using survey data from 1,334 Chinese adults, we combined the Health Belief Model and the Unified Theory of Acceptance and Use of Technology within an explainable machine-learning framework to identify important predictors and their non-linear interactions. Results: The machine learning based optimal XGBoost model (R2=0.647) significantly outperforms traditional regression approaches. Perceived barriers, self efficacy, intention to use smart tools and social support emerge as the four core predictors of the PA I-B gap. Higher levels of perceived barriers and late night frequency enlarge the gap whereas greater self efficacy, perceived exercise benefits, intention to use smart tools, social support, social influence and personal innovation narrow it. The psychological cognition dimension exhibits significantly stronger predictive power than smart sports tools. These tools function primarily as auxiliary resources, and their facilitative effects differ across distinct psychological cognition levels. Conclusion: Psychological cognition and smart sports tools jointly predict the PA I-B gap. The study’s conclusions are constrained by its reliance on self-reported measures and its cross-sectional design. Future research should adopt longitudinal or experimental protocols, supplemented by objective data from wearable devices, to delineate causal pathways and illuminate the finer mechanisms underlying the gap.