Machine learning based identification of key predictors and socioeconomic inequalities in the co-existence of diabetes and hypertension among Bangladeshi adults

基于机器学习的孟加拉国成年人糖尿病和高血压共存关键预测因素及社会经济不平等现象识别

阅读:4

Abstract

Type 2 diabetes mellitus and hypertension often coexist, posing significant public health challenges worldwide. This study aims to identify the associated factors of coexistence of hypertension and diabetes (CHD) among adults in Bangladesh using machine learning (ML) algorithms to rank these predictors according to their relative influence. This was a cross sectional household study. The secondary data was extracted from nationally representative dataset collected by Bangladesh Demographic and Health Survey (BDHS), 2022. We analyzed Bangladeshi 13,541 adults (6,211 males; 7,330 females) aged ≥ 18 years. Chi-square tests and logistic regression identified associated factors, while XGBoost with SHAP quantified feature contributions. Inequalities were assessed using concentration curves (CC), concentration index (CCI), and decomposition analysis. Prevalence was 28.5% for hypertension, 16.6% for diabetes, and 6.9% for CHD. Higher risk of CHD was linked to Chattogram, Dhaka, Sylhet divisions, female sex, urban residence, hygienic toilet use, college education, higher wealth, malnutrition (over/underweight), age ≥ 35 years, and unsafe water use (p < 0.05). SHAP ranked older age as the strongest predictor, followed by BMI, wealth, and region; larger family size was protective. The CC lay below equality (CCI = 0.2651, p < 0.001), showing disproportionate prevalence among the affluent. Wealth (11.7%), BMI (6.1%), region (2.7%), education (2.0%), and residence (1.3%) were key contributors to inequality. CHD in Bangladesh is concentrated among wealthier adults, largely driven by older age, high BMI, and socioeconomic disparities. Addressing these determinants is crucial to reduce the burden and inequality of these conditions.

特别声明

1、本页面内容包含部分的内容是基于公开信息的合理引用;引用内容仅为补充信息,不代表本站立场。

2、若认为本页面引用内容涉及侵权,请及时与本站联系,我们将第一时间处理。

3、其他媒体/个人如需使用本页面原创内容,需注明“来源:[生知库]”并获得授权;使用引用内容的,需自行联系原作者获得许可。

4、投稿及合作请联系:info@biocloudy.com。