Abstract
OBJECTIVE: To investigate the effects of different intensities of physical activity (PA) on cognitive function in middle-aged and elderly individuals, and to predict future trends in cognitive ability using longitudinal data to assess the long-term role of PA in cognitive preservation. METHODS: Data from the China Health and Retirement Longitudinal Study (CHARLS) were utilized. Mixed-effects models were employed to analyze the impacts of low-intensity PA (LPA), moderate-intensity PA (MPA), and vigorous-intensity PA (VPA) on overall cognition, episodic memory, and mental intactness. Random forest and XGBoost machine learning methods were employed to further validate the effects of PA. ARIMA models predicted future cognitive trends under the influence of PA. RESULTS: MPA demonstrated significant advantages in preserving cognitive function, particularly in overall cognition and episodic memory. While LPA had some protective effects, they were less significant than those of MPA, and VPA did not show advantages. Machine learning methods confirmed these findings. ARIMA model predictions indicated that the protective effects of MPA on cognitive function are likely to persist in the future. CONCLUSIONS: Moderate-intensity physical activity is associated with the preservation of cognitive ability in middle-aged and elderly individuals and may continue to provide this benefit in the future; however, further in-depth research is needed for confirmation.