The performance of polynomial ordinal logistic regression analysis on hypertension risk modeling

Abstract

Hypertension prevalence in Indonesia is on the rise. The risk of hypertension exhibits a non-linear relationship with predictors such as age, where the risk increases with age but declines in older individuals. This study aimed to analyze the factors influencing hypertension risk using polynomial ordinal logistic regression analysis. Cholesterol levels and mean arterial pressure were identified as significant linear predictors of hypertension risk. Additionally, age showed a significant polynomial effect, achieving a classification accuracy of 76.7%. The polynomial ordinal logistic regression model demonstrated improved classification accuracy compared to the linear model, increasing from 67.8% to 76.7%. These findings highlight the importance of incorporating non-linear relationships in predictive models to enhance the accuracy of hypertension risk assessment.