Background: Cognitive decline is an escalating public health challenge with global population aging. Understanding the association between hobby engagement and cognitive function is crucial for developing effective interventions. Methods: Utilizing data from waves 2 to 9 of the English Longitudinal Study of Ageing (ELSA), this study included 6854 community-dwelling adults aged 50 years and older. To comprehensively capture this relationship, we employed a multi-method analytical approach, encompassing multiple linear regression, linear mixed-effects models, group-based trajectory modeling (GBTM), and a comparison of machine learning algorithms to assess both associations and predictive performance. Results: Cross-sectionally, hobby engagement was significantly associated with higher baseline global cognitive, memory, and executive function scores. Longitudinally, hobbies were associated with a slower rate of decline in orientation, but not in memory, executive function, or global cognition. Group-based trajectory modeling, which identifies distinct subgroups with heterogeneous cognitive pathways, revealed that hobby engagement was associated with a substantially lower risk of belonging to a “persistently low cognitive function” trajectory (OR = 0.46). Among predictive models, the Scikit-learn Gradient Boosting Regressor demonstrated superior performance (RMSE = 0.7517, R2 = 0.3033), outperforming deep learning approaches. Conclusions: Our findings suggest that hobby engagement may have domain-specific protective effects on cognitive health in older adults, most notably by drastically reducing the risk of a severely adverse cognitive trajectory. Promoting hobby participation therefore emerges as a potential viable, low-cost, and impactful public health strategy for preserving cognitive health in aging populations.
The hemorrhagic lesion segmentation plays a critical role in ophthalmic diagnosis, directly influencing early disease detection, treatment planning, and therapeutic efficacy evaluation. However, the task faces significant challenges due to lesion morphological variability, indistinct boundaries, and low contrast with background tissues. To improve diagnostic accuracy and treatment outcomes, developing advanced segmentation techniques remains imperative. This paper proposes an adversarial learning-based dynamic architecture adjustment approach that integrates hierarchical U-shaped encoder-decoder, residual blocks, attention mechanisms, and ASPP modules. By dynamically optimizing feature fusion, our method enhances segmentation performance. Experimental results demonstrate a Dice coefficient of 0.6802, IoU of 0.5602, Recall of 0.766, Precision of 0.6525, and Accuracy of 0.9955, effectively addressing the challenges in fundus image hemorrhage segmentation.[* Corresponding author.]