PL
The goal of this paper is to apply Generalized Additive Models to medical scheme data. The flexibility of the nonparametric approach is demonstrated based on a real-life empirical example that seeks to model hypertension and the interplay of determinants, such as physiological measurements, medical attributes, demographic and socioeconomic characteristics in predicting blood pressure. The assessment of nonlinear patterns in the response-predictor relationship and the strength of this association are investigated. The extended Generalized Additive Models allow for modeling not only location and scale, but also other distribution parameters, such as kurtosis and skewness