Global emergence and γ-aminobutyric acid type A (GABA(A)) receptor activity of the new designer benzodiazepine ethylbromazolam.

Diabetes milletus (DM) is a major non-communicable disease and a leading contributor to global morbidity and mortality. It is characterized by chronic hyperglycemia and is associated with severe long-term complications, including cardiovascular disease, kidney failure, neuropathy, stroke, and vision impairment. The growing burden of DM poses substantial challenges to health systems worldwide, particularly in low- and middle-income countries (LMICs), where healthcare resources are often limited. Globally, the prevalence of diabetes has risen sharply over recent decades. According to the WHO, the number of individuals living with diabetes increased from approximately 200 million in 1990 to over 800 million in 2022. Similarly, the International Diabetes Federation (IDF) reports that nearly one in ten adults worldwide is affected by diabetes, with projections indicating continued growth. The burden is particularly pronounced in Asia, driven by rapid urbanization, population aging, and lifestyle transitions. In Bangladesh, diabetes has emerged as a critical public health concern. National estimates suggest that approximately 8–10% of adults are living with diabetes, with a significant proportion remaining undiagnosed. Rural and semi-urban populations are especially vulnerable due to limited access to healthcare, low health literacy, and insufficient awareness of preventive behaviors. These disparities underscore the need for region-specific investigations to better understand the epidemiology and determinants of T2D. Previous studies in Bangladesh and elsewhere have primarily relied on traditional statistical approaches, such as logistic regression, to estimate prevalence and identify risk factors. While these methods are valuable, they may be limited in capturing complex, nonlinear relationships among predictors. In recent years, machine learning (ML) techniques including decision trees, random forests, support vector machines, and gradient boosting have been increasingly applied to diabetes prediction, demonstrating improved predictive performance. Recent studies have also incorporated explainable artificial intelligence (XAI) techniques such as SHAP to improve model interpretability and identify clinically relevant risk factors. Several studies have reported strong predictive accuracy using ML models. For instance, artificial neural networks have achieved accuracy up to 88.6%, while -nearest neighbor models have demonstrated accuracy exceeding 96% with high sensitivity and specificity. Advanced approaches such as generative adversarial networks and ensemble tree-based models have further improved classification performance, achieving area under the curve (AUC) values as high as 0.97. Despite these advancements, most existing studies have focused on single-model approaches, with limited exploration of EML techniques that combine multiple models to enhance predictive robustness. Moreover, to the best of our knowledge, no prior community-based study in northern Bangladesh has integrated stacking-based EML, SHAP-based explainability and traditional multivariable logistic regression for diabetes prediction and risk factor identification using real-world population-level data. This represents a critical research gap, as ensemble approaches are particularly effective in improving generalizability and capturing complex interactions among socio-demographic and behavioral variables. The northern region, particularly Dinajpur district, is characterized by distinct socio-demographic and economic challenges, including lower educational attainment, limited healthcare infrastructure, and suboptimal lifestyle practices, which may contribute to an elevated risk of DM. However, comprehensive community-based studies integrating advanced analytical techniques in this region remain scarce. Therefore, this study aims to predict the DM among adults in northern Bangladesh, identify key socio-demographic and behavioral risk factors using EML approaches. Then finally, validate these findings through traditional multivariable logistic regression analysis. By integrating advanced predictive modeling with conventional statistical approaches, this study seeks to generate robust, interpretable, and generalizable evidence for diabetes risk stratification. The findings of this study have important implications for public health policy and practice. They can inform targeted screening, early detection, and prevention strategies, particularly among underserved populations. Furthermore, this research aligns with the United Nations Sustainable Development Goals (SDG 3: Good Health and Well-being), contributing to efforts aimed at reducing premature mortality from non-communicable diseases through evidence-based and data-driven interventions in Bangladesh. A total of 1408 adults participated in the study, with the majority residing in rural areas (77.9%) and a slightly higher proportion of males (52.3%) than females (Table ). Most participants wer