Hypertension risk modeling using penalized spline estimator approach based on consumption of salt, sugar, and fat factors

Hypertension is one of the health problems that arise without symptoms. It means that emergence factors of symptoms of hypertension cannot be known certainty. There were some previous researchers who pointed out that the risk factors of hypertension can be caused by overweight (obese), heredity, age, and history of family’s life. However, according to study run by World Health Organization (WHO) in 2013, risk factors of hypertension may also be caused by the consumption of fatty acids, saturated fat, salt and sugar. In that research, WHO used parametric regression model approach. So, this model cannot accommodate locally behavior of risk factors. Therefore, in this research, we propose analysis methods by using both parametric logistic regression and nonparametric logistic regression models approaches that can accommodate locally behavior of risk factors. In this research, to model hypertension caused by consumption of salt, sugar, and fat we use link function of logit in parametric logistic regression and use penalized spline estimator in nonparametric logistic regression. The results show that classification accuracy based on nonparametric logistic regression is 90.7% and based on parametric logistic regression is 65.6%. It means that for modeling the risk of hypertension based on consumption of salt, sugar, and fat, the use of penalized spline estimator of nonparametric logistic regression is better than that of logit link function of parametric logistic regression.