Data Analysis of Diabetes Mellitus with Joint Modeling Method

Diabetes mellitus is a dangerous disease that requires long-term medical treatment. The cause of this disease is high blood sugar levels. If not treated immediately, complications will occur and even cause death. The data is taken from the Indonesia Family Life Survey (IFLS). IFLS is a longitudinal measurement that is performed repeatedly every five years. More data is needed for repeated measures. Therefore, this research needs to be done to accommodate the missing data, and it is assumed that it is missing at random (MAR). This study aims to analyze the causative factors that are thought to affect the recovery time of patients with diabetes mellitus using the joint modeling method. This model is a relationship between event time data and repeated measurement data. The joint modeling method uses a linear mixed model for longitudinal measurements and a Cox proportional hazard model for survival. The variables were taken from IFLS4 and IFLS5 data with 293 observations: measurement time, treatment history, gender, comorbidities, and complications. The results in this study obtained a significant influence, namely the variables of measurement time, gender, and complications, on the recovery time of patients with diabetes mellitus. With the reduced measurement time, the patient has a lower chance of recovering 8.7184 times. The variables of gender also have a lower possibility of recovery of 9.1032 times, respectively.