NATIONAL HEALTH INSURANCE (JKN) PARTICIPANTS CONTRIBUTION CLASS AND HEALTH FACILITIES BILLING FEES MODELING USING GENERALIZED LINEAR MODEL (GLM)
Health is one of the main problems in managing finances for each individual. Health insurance companies need to pay attention to the factors that affect the amount of health costs. In Indonesia, there is a scheme established by the Social Security Administrator (BPJS) for Health, namely the National...
Saved in:
| Main Author: | |
|---|---|
| Format: | Final Project |
| Language: | Indonesia |
| Online Access: | https://digilib.itb.ac.id/gdl/view/72995 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Institut Teknologi Bandung |
| Language: | Indonesia |
| Summary: | Health is one of the main problems in managing finances for each individual. Health insurance companies need to pay attention to the factors that affect the amount of health costs. In Indonesia, there is a scheme established by the Social Security Administrator (BPJS) for Health, namely the National Health Insurance (JKN). This scheme aims to help all Indonesian people, especially the poor, to obtain health insurance services with affordable contributions. Even though this scheme has helped underprivileged communities in Indonesia, JKN has experienced a deficit because the fees billed from health facilities are far greater than the premiums paid. It is necessary to analyze the factors that affect the class contributions of participants and the fees billed from health facilities using the Generalized Linear Model (GLM). Participation data is used to model the participant contribution class as a response variable, with the predictor variables used are: gender, province of residence, segmentation, and age of participants. The contribution class is considered to have an ordinal logistic distribution because it has a different level for each class. FKRTL service data is used to model FKRTL billing fees to BPJS Health as a response variable, with predictor variables used are: level of service, regional rates, province of FKRTL, class of participant contributions, disease severity, and disease cases. FKRTL billing fees to BPJS Health are considered to have an Inverse Gaussian distribution taking into account the residual deviation. The best model for the participant contribution class has a misclassification error value of 19,5652%, with the most influential predictors are: participant segmentation, participant's province of residence, and gender. The best model for billing fees for FKRTL to BPJS Health has a MAPE value of 48,1861%, with the most influential predictors namely: disease cases, class contributions of participants, and regional rates of FKRTL. |
|---|