Stacking-based multi-objective ensemble framework for prediction of hypertension

Hypertension is a common health problem that is costly to treat, difficult to control, and frequently results in serious and fatal disorders like cardiovascular disease (CVD) and stroke. The main objective of this work was to design and verify a stacking ensemble framework-based model for predicting hypertension risk prospectively. Firstly, we proposed a Multi-objective Iterative Model Selection (MoItMS) strategy to maximize the accuracy of meta-learners and the diversity of the ensemble model at the same time. An effective method for classifying people for managing population health and assisting in the assessment and identification of hypertension is then provided using a stacking-based multi-objective ensemble framework that can be applied to enormous volumes of clinical data. The National Health and Nutrition Examination Survey (NHANES) collected data from 2007 to 2018. Of the 11,341 patients studied, 67.16 % were non-hypertensive and 32.84% were hypertensive, resulting in an imbalanced data set. According to the findings, the model outperformed 13 individual models and ensemble models in terms of precision (71.13 %), recall (53.76 %), accuracy (76.82 %), F1-measure (61.05 %) and AUC (area under the curve) of 0.84. Furthermore, the proposed ensemble framework produced results that were somewhat more successful (AUC = 0.788) than prior hypertension research using an artificial neural network with similar input features, which produced an AUC of 0.77. We focused on the impact of lifestyle factors on hypertension classification performance and discovered that lifestyle factors can improve the model in distinguishing hypertensive samples. Identifying people at high risk of hypertension will be easier with our method, which we hope to integrate into community health management systems in the future.