Identifying Gender-Specific Determinants of Diabetes in Sudan Using Statistical and Machine Learning Techniques: Evidence from the WHO STEPwise Survey

Background Diabetes mellitus is a rapidly growing global public health concern, particularly in low- and middle-income countries. Sudan is experiencing a rising burden of non-communicable diseases while still facing infectious diseases, placing considerable strain on its healthcare system. This study estimates the prevalence of diabetes and identifies associated risk factors among Sudanese adults using the 2016 WHO STEPwise survey. Methods The analysis included 7,819 Sudanese adults aged 18–69 years. Explanatory variables included demographic factors (age and gender), anthropometric indicators (body mass index and waist circumference), clinical measures (blood lipids and systolic blood pressure), and socioeconomic characteristics (income and occupation). Descriptive statistics and chi-square tests were used for preliminary analysis. Quantile regression, multinomial logistic regression, and nonparametric kernel regression were applied to identify determinants of diabetes. Additionally, a Random Forest model with Leave-One-Covariate-Out (LOCO) analysis was used to assess the relative importance of predictors. Results The prevalence of diabetes among Sudanese adults was 9.7%, with slightly higher prevalence among women (9.7%) than men (9.5%). Age was a significant determinant for both sexes, with individuals aged 45–59 showing a substantially higher risk (OR = 2.47, p < 0.001). Obesity significantly increased the likelihood of diabetes among women (OR = 3.14, p < 0.001). Elevated blood lipid levels were strongly associated with diabetes among both men (OR = 11.00, p < 0.001) and women (OR = 4.68, p < 0.001). High blood pressure also increased diabetes risk among women (OR = 4.71, p < 0.001 for BP ≥ 160 mmHg). Nonparametric smoothing curves indicated age-related increases in diabetes prevalence, although the patterns differed by gender. Random Forest analysis identified BMI as the most influential predictor, followed by blood pressure, income, and age. Conclusion Diabetes among Sudanese adults is influenced by demographic, clinical, and socioeconomic factors, with notable gender differences. Public health interventions targeting modifiable risk factors—particularly obesity and hypertension—are essential for reducing the growing diabetes burden in Sudan.