Key result
Hybrid XGBoost-SHAP model predicts hypertension with ~66% accuracy, identifying BMI and age as top drivers.
Why the study?
Hypertension is a chronic disease with high prevalence that often shows no symptoms at an early stage, motivating an approach to improve the accuracy and interpretation of hypertension risk prediction.
Does a hybrid XGBoost and SHAP machine learning model accurately predict and interpret the risk of prevalent hypertension in a secondary dataset of 1,007 individuals?
Observational (n=1,007)
Does a hybrid XGBoost and SHAP machine learning model accurately predict and interpret the risk of prevalent hypertension in a secondary dataset of 1,007 individuals?
Combining XGBoost with SMOTE and SHAP provides a moderately accurate (AUC 0.707) and interpretable framework for early hypertension risk screening, identifying BMI and age as the most influential predictors.
Answers are grounded in this paper and cite their sources.
Moderate accuracy limits clinical adoption; hypothesis-generating for interpretable ML hypertension models pending prospective validation.
This study proposed a machine learning-based approach to improve the accuracy and interpretation of hypertension risk, which is a chronic disease with high prevalence and often shows no symptoms at an early stage. The developed model combined the Extreme Gradient Boosting (XGBoost) algorithm with an explainable AI method, Shapley Additive Explanations (SHAP), to produce predictions that were not only accurate but also understandable. The data used in this study was a secondary dataset based on demographic and clinical characteristics, which was processed through preprocessing and class imbalance handling using SMOTE. The evaluation results showed that the model achieved an accuracy of 66.33% with an AUC of 0.707, indicating a moderate classification performance. In addition, the recall value for the hypertension category was 0.61, showing that the model was fairly effective in detecting at-risk individuals. SHAP analysis found that body mass index and age were the main factors in predicting hypertension. This study emphasized that combining predictive and interpretable models could be an effective approach in developing decision support systems in the healthcare sector.
No takes yet. Share an insight, caveat, or question.
Manurung et al. (2026) conducted an observational in Hypertension (n=1,007). Hybrid XGBoost and SHAP risk prediction model (evaluating BMI, age, heart rate, smoking status, and education) was evaluated on Hypertension risk classification performance (accuracy, AUC, and recall) (AUC 0.707, Accuracy 66.33%, Recall 0.61). A hybrid XGBoost-SHAP model predicted hypertension with 66.33% accuracy and an AUC of 0.707, identifying BMI and age as the primary risk predictors.
Synapse has enriched 10 related papers. Consider them for comparative context: