Background: We aimed to examine the association between sleep duration and cognitive impairment among elderly Chinese people. Methods: generalized linear modeling was used to analyze the baseline data for adults aged 65 years and over (n = 4785) selected from the 2011 China Health and Retirement Longitudinal Study (CHARLS). The two aspects of cognitive impairment assessed in the study were mental status and memory. Sex-stratified logistic regression models were conducted to identify the effect of sleep duration in the different gender groups. Results: of all the participants, 49.8% were females and 32.5% aged 75 years and over. Of the participants, 59.7% had short sleep duration (8 h/day). Compared to normal sleep duration, long sleep duration was associated with lower mental status scores (β = −0.43, p = 0.001) and lower memory scores (β = −0.26, p = 0.006). Long sleep duration was associated with lower mental status in both men (β = −0.37, p = 0.033) and women (β = −0.46, p = 0.025), however, only in men was long sleep duration found to be associated with low memory scores (β = −0.26, p = 0.047). Conclusions: Our study showed that long sleep duration was significantly associated with poorer mental status and memory scores in elderly Chinese people. Paying greater attention to the effects of sleep patterns on the risk of cognitive decline may yield practical implications for dementia prevention and health promotion, especially among older women with lower educational attainment, living in rural areas, and those who have long sleep duration.

We propose a Bayesian kernel machine regression (BKMR) framework for count outcomes with dynamic spatiotemporal dependence. The proposed model, termed Negative Binomial BKMR with spatiotemporal effects (NB-BKMR), integrates (i) a negative binomial likelihood to accommodate overdispersion, (ii) a kernel-based exposure–response surface for complex mixtures, (iii) hierarchical group-wise variable selection and (iv) a dynamic spatiotemporal random effect structure based on a Leroux conditional autoregressive (CAR) prior evolving over time. Posterior inference is conducted in a fully Bayesian framework using Polya-Gamma data augmentation. Through simulation studies, under varying nonlinear exposure–response functions, correlation structures, and spatiotemporal dependence patterns, we show that NB-BKMR yields well-calibrated uncertainty quantification and robust identification of dominant mixture drivers, even when exposures are highly correlated. An application to the U.S. state-level traffic fatality counts (1982–1988) illustrates how the model uncovers nonlinear effects and interactions among socioeconomic and behavioral predictors while improving predictive performance relative to generalized additive models with spatiotemporal smooths. This work extends existing BKMR methodology by unifying mixture modeling, count outcomes, and dynamic spatial dependence in a single coherent framework, with particular relevance for areal public health surveillance data.