Machine learning classification analysis for a hypertensive population as a function of several risk factors

摘要

This research presents a prediction model to evaluate the association between gender, race, BMI, age, smoking, kidney disease and diabetes using logistic regression. Data collected from NHANES datasets from 2007 to 2016. An unbalanced sampling dataset of 19.709 with (83%) non-hypertensive individuals and (17%) hypertensive individuals. Some risk factors were categorized, and indicator variables were created to transform the continuous variables to a binary form to have consistent predictors with the outcome. The results show a sensitivity of 77%, a specificity of 68%, precision on the positive predicted value of 32% in the test sample and a calculated AUC of 73% (95% CI[0.70-0.76]). The model also confirms that individuals with obesity, age range between 71 and 80 years old, race non-Hispanic black and male have higher odds of having hypertension. Diabetes, kidney disease and smoking habits do not affect odds of the outcome. In clinical practice, this model can be used to inform patients and guide population health management in detecting patients with high probability of developing a cardiovascular disease. The proposed logistic regression method can be used as an expert system's inference engine to support the experts in the cardiovascular disease field to provide problem analysis for patients in risk of developing hypertension. (C) 2018 Elsevier Ltd. All rights reserved.