An integrated machine learning framework for the early diagnosis of hypertension disease

• Published in: Engineering applications of artificial intelligence Vol. 162; p. 112564
• Main Author: Eldem, Ayşe
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 22.12.2025
• Subjects: Extreme gradient boosting; Feature selection; Framework; Hypertension disease; Machine learning; Hypertension disease; Feature selection; Extreme gradient boosting; Framework; Machine learning
• ISSN: 0952-1976
• DOI: 10.1016/j.engappai.2025.112564

Hypertension is a type of disease that occurs after a serious increase in blood pressure. Due to the rapid increase in the disease, efforts for its early diagnosis are increasing day by day. The use of artificial intelligence (AI) and machine learning (ML) methods in disease detection is of great importance, especially in early diagnosis. In this study, a framework for the diagnosis of hypertension is developed. Behavioral Risk Factor Surveillance System (BRFSS) and Hypertension Risk Prediction (HRP) datasets with clinical and demographic information were used to diagnose hypertension. The problem of class imbalance in the datasets was solved by Random OverSampling method. In particular, the relationship between the attributes and the target variable was analyzed with Consistency-Based Filter, Recursive Feature Elimination with Cross-Validation (RFECV), Least Absolute Shrinkage and Selection Operator (LASSO), Information gain, Relief and Correlation-based feature selection methods and the features selected for the diagnosis of hypertension disease were taken into consideration. In the experiments, Extremely Randomized Trees (Extra Trees), Adaptive Boosting (Adaboost) and Extreme Gradient Boosting (XGBoost) machine learning methods were used in five fold cross-validation. In both datasets, the most successful results were obtained with the XGBoost algorithm. Then, for this algorithm, Random Search, Grid Search, Bayes Search and Genetic Algorithm were used for model hypertuning. For the BRFSS dataset with 87.96 % accuracy and for the HRP dataset with 97.67 % accuracy were obtained. The results are obtained with evaluation metrics that the medical world considers in evaluations, and the proposed framework for hypertension disease detection provides valuable insights for healthcare applications. •Proposes MLHyDD, a framework to diagnose hypertension disease.•Alleviates the imbalanced classification problem and uses different feature selection methods.•Applies hyper-tuning by selecting the most successful machine learning technique with a self-supervised comparative strategy.•Extensive experiments demonstrate both the effectiveness and interpretability of the framework.