Abstract
This study presents a machine learning approach to forecast hypertension within urban inhabitants, focusing on the Rangpur district, Bangladesh for data col- lection and model training. Ten machine learning algorithms, such as Logistic Regression, Gaussian Naive Bayes (GNB), Support Vector Machine (SVM), K- Nearest Neighbor (K-NN), Decision Tree (DT), Random Forest (RF), Bagging, AdaBoost, Gradient Boosting (GB), and Extra Tree (ET) are utilized to enhance the accuracy of predicting hypertension risk in this specific region. Data gath- ered from 611 patients across different healthcare facilities, containing details like blood pressure measurements, medical records, and hypertension diagno- sis, form the dataset for analysis. The aim of this research is to enhance early detection techniques and customize public health interventions in Rangpur City. Examination of the primary data establishes a substantial association between hypertension and blood pressure parameters (0.79 for Sys BP, 0.78 for Dia BP) in comparison to other variables. Evaluation of model performance is based on metrics like accuracy, precision, recall, and F1-score. Findings demonstrate that the AdaBoost model exhibits superior performance indicators, achieving 98.37% accuracy, 100% precision, 96.87% recall, and an F1-score of 98.39% when trained on 70% of the dataset and evaluated on 30%, with a focus on blood pressure. Even excluding this attribute, the AdaBoost model surpasses others with 78.86% accu- racy, 77.14% precision, 84.38% recall, and an F1-score of 78.79% when trained on 80% of the dataset and tested on 20%. By prioritizing early detection and pre- ventive healthcare, Bangladesh’s healthcare system can diminish expenses linked to costly therapies and hospital stays.