Wei Wu

Background: China has experienced substantial dietary transitions over recent decades, accompanied by increasing burdens of overweight and obesity. However, long-term changes in dietary patterns and their associations with obesity among middle-aged and older adults remain insufficiently characterized. Methods: We analyzed data from 8502 adults aged ≥45 years participating in four waves (2002–2022) of the China National Nutrition Surveys in Guangdong Province. Dietary patterns were identified using k-means cluster analysis based on 23 food groups. Temporal changes in dietary pattern distributions were assessed using Rao–Scott adjusted chi-square tests and survey-weighted logistic regression models incorporating linear and quadratic terms for survey year. Multivariable logistic regression models were used to examine associations between dietary patterns and overweight, obesity, and central obesity. Results: Three dietary patterns were identified: a traditional cereal-based pattern, a transitional mixed pattern, and an animal-rich diverse pattern. The prevalence of the Traditional Cereal-based pattern decreased from 15.8% in 2002 to 0.4% in 2022, whereas the Animal-rich Diverse pattern increased from 19.2% to 47.0%. Trend analyses confirmed significant temporal changes in all dietary patterns after false discovery rate correction. Compared with the Traditional Cereal-based Pattern, the Transitional Mixed Pattern and Animal-rich Diverse Pattern was associated with higher odds of overweight/obesity, with adjusted ORs of 1.57 (95% CI: 1.29–1.90) and 1.42 (95% CI: 1.14–1.77), respectively. Similar associations were observed for central obesity, with corresponding adjusted ORs of 1.54 (95% CI: 1.26–1.89) and 1.35 (95% CI: 1.08–1.71), respectively. Conclusions: Significant dietary transitions occurred among middle-aged and older adults in Guangdong Province from 2002 to 2022. Dietary patterns characterized by higher intakes of animal-source and energy-dense foods were associated with greater odds of overweight/obesity and central obesity. These findings suggest that dietary diversity alone may not necessarily represent a healthier dietary pattern during nutrition transition and highlight the potential importance of considering overall dietary composition and energy balance in obesity prevention strategies.

Background: This study aimed to investigate whether the triglyceride-glucose (TyG) index mediates the association between remnant cholesterol (RC) and coronary artery disease (CAD) severity and to evaluate their combined association with multi-vessel CAD. Methods: In this retrospective study, anonymized data from CAD patients undergoing coronary computed tomography angiography (CCTA) or coronary angiography (CAG) between 2019 and 2022 were analyzed . Patients were classified into single-vessel or multi-vessel disease groups. Insulin resistance was estimated using the TyG index, and RC was calculated. Multivariate logistic regression, restricted cubic spline (RCS) modeling, and mediation analysis were utilized to explore the relationships among the TyG index, RC, and CAD severity. Results: A total of 592 patients (64.7% male; mean age, 62 years) were included in the analysis. The TyG index was positively correlated with RC ( r = 0.52, P < 0.001). After adjustment for confounding factors, each 1-SD increase in the TyG index was associated with higher odds of multi-vessel CAD (OR = 1.724; 95% CI, 1.323–2.247; P < 0.001), and each 1-SD increase in the RC was similarly associated with higher odds of multi-vessel CAD (OR = 1.491; 95% CI, 1.024–2.171; P = 0.037). Patients with both high TyG index and high RC had significantly higher odds of multi-vessel CAD (OR = 3.53; 95% CI, 2.040–6.258; P < 0.001). Mediation analysis demonstrated that the TyG index mediated 33% of the association between RC and multi-vessel CAD. Conclusions: The TyG index is positively correlated with RC and is significantly associated with CAD severity. Furthermore, the TyG index partially mediates the association between RC and CAD severity. Combined assessment of the TyG index and RC may improve metabolic risk stratification in patients with CAD.

Abdominal obesity (AO) significantly contributes to cardiometabolic diseases and poses an increasing public health challenge. Composite lipid-derived indices have been proposed as simple indicators of atherogenic dyslipidemia, but their relative importance for identifying AO at the population level remains incompletely understood. This study aimed to evaluate the link between various atherogenic lipid indices and AO, identifying key lipid-related predictors through interpretable machine learning methods. This cross-sectional study analyzed 5,612 adults from a population-based survey in Guangdong Province, China. Atherogenic lipid indices encompass the atherogenic index of plasma (AIP), non-high-density lipoprotein cholesterol (non-HDL-C), atherogenic coefficient (AC), Castelli risk indices I and II (CRI-I and CRI-II), lipid composite index (LCI), and the triglyceride-to-high-density lipoprotein cholesterol ratio (TG/HDL-C). Feature selection was conducted using Boruta and Least Absolute Shrinkage and Selection Operator (LASSO) methods prior to machine learning modeling. Model performance was evaluated using the area under the receiver operating characteristic curve (AUC), accuracy, and F1 score. SHapley Additive exPlanations (SHAP) analysis was applied to quantify feature importance. Individuals with AO were generally older and more likely to exhibit lower education levels, reduced physical activity, cardiometabolic comorbidities, and unfavorable metabolic profiles. Feature selection identified 15 key predictors. Logistic regression demonstrated the most stable performance (training AUC: 0.767; testing AUC: 0.768; accuracy: 0.712; F1 score: 0.667) with good calibration and clinical utility. SHAP analysis consistently identified AIP, sex, CRI-II, uric acid, and diastolic blood pressure as the most influential predictors of AO. Composite lipid indices, particularly AIP and CRI-II, are strongly associated with AO and may serve as practical indicators for identifying individuals at elevated metabolic risk. Because these indices are derived from routinely measured lipid parameters, they may support scalable approaches for metabolic risk screening and monitoring in both clinical and community health settings.

Metabolic syndrome (MetS) is a cluster of risk factors for cardiovascular diseases and type 2 diabetes. Gut microbiota dysbiosis has been implicated in the pathogenesis of MetS, but the mechanisms remain poorly understood. This study investigates gut microbiota interaction networks in MetS and explores their potential role in host metabolic regulation. In this population-based cross-sectional study, 221 MetS patients and 382 healthy controls were analyzed. Co-abundance network analysis was used to examine microbial interactions across the study. Significant differences in microbial co-abundance patterns were observed between MetS and healthy participants. In MetS, the gut microbiota displayed fewer and generally weaker co-abundance correlations compared with healthy controls. These changes appear to be more strongly associated with the synergistic effects of microbial interactions than solely with the abundance of individual taxa investigated here. Specific microbiota combinations were found to influence key metabolic functions, contributing to MetS development. The findings suggest that microbial interactions, rather than the abundance of individual bacteria, are associated with MetS. This study provides new insights into the role of disrupted gut microbiota networks in MetS pathogenesis.

Background: Metabolic syndrome (MetS) is a rising public health concern in Southern China, with limited evidence available on dietary and metabolic factors. This cross-sectional study employed an interpretable machine learning (ML) approach to identify factors that could inform clinical and community interventions. Methods: Data were obtained from the Guangdong Nutrition Surveys conducted in 2015 and 2022, including sociodemographic information, lifestyle patterns, physical examinations, laboratory measurements and dietary intake information (collected via repeated 24-h dietary recalls). Potentially relevant variables were selected using the Least Absolute Shrinkage and Selection Operator (LASSO) regression and incorporated into seven ML models. Model performance was primarily assessed using the area under the receiver operating characteristic curve (AUC), and the contribution of identified features was interpreted through SHapley Additive exPlanations (SHAP). Results: This analysis included 5593 participants, of whom 1103 were classified as having MetS. After removing collinear features, the ML models retained 19 candidate variables, which were selected through LASSO regression. XGBoost achieved the best performance (AUC: 0.834; F1 score: 0.537) with a misclassification rate of 27.1%. SHAP analysis highlighted body mass index (BMI), age, and uric acid levels as major risk factors, while insoluble dietary fiber, carbohydrate and specific micronutrients exhibited protective associations. Conclusions: Machine learning identified key dietary and metabolic factors of MetS. Integrating these factors into clinical practice and public health initiatives may enhance early detection and support personalized prevention strategies for MetS in Southern China.

Atopic dermatitis (AD) is characterized by chronic and recurrent itching with a high burden of disability-adjusted life years (DALYs, a measure of overall disease burden). Traditional treatments mainly include corticosteroids, which have a good effect on controlling inflammation but adverse side effects. Recently, advancements in understanding the pathogenesis of AD have led to the emergence of a variety of novel therapeutic approaches, such as microbiome manipulation, offering renewed hope for more effective management of this condition. These strategies are particularly promising for mild-to-moderate AD, where dysbiosis and immune imbalance (e.g., Th2 skewing) are key drivers, though some approaches (e.g., fecal microbiota transplantation) are being explored for refractory cases. It has been shown that microbiome manipulation has the potential to improve disease states and regulates the balance of the inflammatory system in a variety of ways. Various approaches have been preclinically and clinically tested, including probiotics (and multiple co-applications), prebiotics, postbiotics, unmethylated CpG motifs, fecal microbiota transplantation, herbal fermentation technology with microorganisms and phage. In this review, we discuss these microbiome manipulation methods and emphasizes the potential of microbiome-based interventions to modulate Th1/Th2 balance with fewer side effects, ultimately leading to control of inflammation in AD. Further translational research in this field is needed to integrate when we apply this therapy and the capability for disease treatment and prevention.

Background: The “gut–skin axis” has been proposed to play an important role in the development and symptoms of atopic dermatitis. Therefore, we have constructed an interpretable machine learning framework to quantitatively screen key gut flora.Methods: The 16S rRNA dataset, after applying the centered log-ratio transformation, was analyzed using five different machine learning models: random forest, light gradient boosting machine, extreme gradient boosting, support vector machine with radial kernel, and logistic regression. Interpretable machine learning methods, such as SHAP values, were used to identify significant features associated with atopic dermatitis.Results: Random forest performed better than the other “tree” models in the validation partitions. The SHAP global dependency plot indicated that Bifidobacterium ranked as the strongest predictive factor across all prediction horizons, although the SHAP values for some features were still higher in support vector machine and logistic regression models. The SHAP partial dependency plot for “tree” models showed that the best segmentation point for Bifidobacterium was further from the origin compared to other features in the respective models, quantitatively reflecting differences in gut microbiota.Conclusion: Machine learning models combined with SHAP could be used to quantitatively screen key gut flora in atopic dermatitis patients, providing doctors with an intuitive understanding of 16S rRNA sequencing data to support precision medicine in care and recovery.

Human papillomavirus (HPV) infection remains a critical public health challenge in China, particularly in Guangdong Province, where HPV-52, 16, and 58 genotypes predominate, and male infection rates exceed 40%. Despite the successful implementation of a government-funded school-based program that has achieved 88% HPV vaccine coverage among adolescent girls, several persistent barriers, including genotype mismatch (the free HPV vaccine covers < 50% of high-risk local strains), regional disparities (80% vs. 60% for first-dose coverage), and exclusion of males, thwart progress toward herd immunity. Financial sustainability risks pose an even more significant threat to the expansion of HPV vaccination programs, especially in Guangdong province where annual expenditures exceed CNY 200 million. This review delves into Guangdong’s pioneering efforts and proposes practical solutions: accelerating domestic multivalent HPV vaccine development, adopting gender-neutral vaccination policies, and leveraging mobile clinics for remote populations. These strategies not only provide a roadmap for China but also serve as valuable insight for other LMICs striving to overcome HPV-related inequalities.

Cardiometabolic health is closely linked to diet. Both specific food choices and dietary indices are commonly used to guide cardiometabolic protection strategies. However, the balance between foods may also significantly impact cardiometabolic health. Here, we explore the balance and imbalance in long-term dietary food intake to quantify food pairing patterns (FPs), providing insights beyond conventional dietary indices and single-food intake frequencies. Using data from our GGMP (n = 6,994) and the NHANES (n = 7,350) cohorts, we observed that long-term food pairing patterns are independent of single food intake frequency and dietary indices. We identified 1,759 and 306 cardiometabolic-related long-term food pairing patterns from two cohorts (FDR < 0.05), respectively. Notably, around 80% of these pairing foods not individually associated with cardiometabolic traits, and food pairing pattern associations with cardiometabolic traits at hyper food group level were consistent across Eastern and Western populations. Besides, mediation analysis revealed that 72.7% of long-term food pairing patterns affected cardiometabolic traits through 31 microbial genera, with Clostridiumsensustricto1 playing a predominant role. Moreover, multi-step mediation analysis found that microbes mediated the impact of long-term food pairing patterns on cardiometabolic traits primarily through their metabolic pathways, such as pyruvate fermentation to propanoate and ergothioneine biosynthesis pathways. Our data suggest that balance between dietary foods is broadly associated with cardiometabolic traits by modulating gut microbial functionalities, in contrast to the lower occurrence when considering individual foods in isolation. These results offer a novel perspective for designing personalized dietary strategies beyond present dietary indices to enhance cardiometabolic health.

The gut microbiome, recognized as a critical component in the development of chronic diseases and aging processes, constitutes a promising approach for predicting host health status. Previous research has underscored the potential of microbiome-based predictions, and the rapid advancements of machine learning techniques have introduced new opportunities for exploiting microbiome data. To predict various host nonhealthy conditions, this study proposed an integrated machine learning-based estimation pipeline of Gut Age Index (GAI) by establishing a health aging baseline with the gut microbiome data from healthy individuals. We assessed the performance of GAI pipeline on two extensive cohorts - the Guangdong Gut Microbiome Project (GGMP) and the American Gut Project (AGP). In the GGMP cohort, for 20 common chronic diseases such as metabolic syndrome, obesity, and cardiovascular diseases, the proposed GAI achieved a balanced accuracy, ranging from 66 to 75%, with the prediction performance for atherosclerosis being the highest. In the AGP cohort, the balanced accuracy of GAI ranged from 58 to 72% for 10 diseases. Based on the results from these two datasets, we conclude that our proposed approach in this study can be used to predict individual health status, which offers the potential for scalable, cost-effective, and personalized health insights.