Machine Learning Approaches for Estimating Prevalence of Undiagnosed Hypertension among Bangladeshi Adults: Evidence from a Nationwide Survey

Abstract

In South Asia, hypertension is the most prevalent modifiable risk factor for cardiovascular disorders. Comparing machine learning to statistical approaches, it has been found that it performs better at identifying clinical risk. This study utilized machine learning techniques to estimate undiagnosed hypertension. We created a single dataset out of individual-level data from the Bangladesh Demographic and Health Survey (2017-18). The JNC-7 and ACCAHA criteria were used to define hypertension. We used two well-known ML approaches logistic regression and log-binomial regression to determine the prevalence of undiagnosed hypertension. A considerable number (16%) of hypertension cases in Bangladesh are still undiagnosed. Young people and the divisions of Sylhet and Rangpur were found to be more at risk for undetected hypertension. ML models performed well at identifying undiagnosed hypertension and its contributing factors in South Asia. Future studies incorporating biological markers will be necessary to improve the ML algorithms and determine their applicability.