|
|
| |
| Predicting of Diabetes Mellitus Type 2 Risk Using Nonparametric Ordinal Logistic Regression Based on Smoothing Spline Estimator
Download PDF
|
Sakinah Priandi, Nur Chamidah, Sediono, Toha Saifudin, M. Fariz Fadillah Mardianto
|
Abstract:
Diabetes mellitus is one of the deadliest diseases in the world. Indonesia ranks seventh for the highest diabetes sufferers in the world. The prevalence of diabetics in Indonesia reaches 6.2%. According to these results, most diabetics suffer from diabetes mellitus type 2. Dyslipidemia is one of the factors that affect diabetes mellitus type 2. Dyslipidemia is a disorder of fat metabolism which is characterized by an increase in plasma fat levels. The main abnormality of fat levels is an increase in LDL cholesterol and triglycerides. To predict the risk of diabetes mellitus type 2, we need to build a model. In statistical analysis, there are two approaches to estimating the model, namely parametric and nonparametric. In this study, we predicted the risk of diabetes mellitus type 2 based on LDL cholesterol and triglyceride levels using nonparametric ordinal logistic regression (GAM) based on a smoothing spline estimator and compared it with the parametric ordinal logistic regression (GLM) approach. Based on stability test, namely the press's Q value based on nonparametric ordinal logistic regression (GAM) model based on the smoothing spline estimator is stable or consistent with the accuracy value and the sensitivity value for the category diabetes mellitus type 2 are 85% and 0.93 respectively. Meanwhile, the Press's Q value based on parametric ordinal logistic regression (GLM) model is instable or inconsistent with accuracy value and the sensitivity value for the category diabetes mellitus type 2 are 37.5% and 0.47 respectively. This means ordinal logistic regression using a nonparametric approach (GAM) based on a smoothing spline estimator is better than a parametric approach (GLM) to predict the risk of diabetes mellitus type 2.
|
|
|
| | |