Truncated spline quantile regression model on platelet changes in dengue fever patients based on body temperature

Abstract

Quantile regression is one of the models used for data containing outliers. The data that has been analyzed by many researchers using quantile regression is dengue fever data. In this study, we propose using quantile regression with truncated spline in modeling platelet data based on the body temperature of dengue patients. We use knot points ranging from 1, 2, to 3 points and the optimal model uses 3 knot points, namely 36, 37.1, and 39.1. In the 0.25 and 0.75 spline quantile models, four patterns of platelet changes were found that tended to be the same. In contrast to the 0.50 spline quantile model, it is seen that platelets actually decrease when the temperature is below 36^oC. However, the three models showed that platelets decreased when the body temperature was high, reaching 39.1^oC.