HEALTH INSURANCE CLAIM MODEL IN JAVA REGION DUE TO THE INFLUENCE OF SOCIAL HEALTH INSURANCE ADMINISTRATION BODY USING THE GENERALIZED ADDITIVE MODEL METHOD

• Main Author: Maharani Tatiana Talaohu, Ainun
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/86931
• Institution: Institut Teknologi Bandung

COVID-19 has had a huge impact on various aspects and fields around the world since 2020. During the pandemic, many people were infected with COVID-19, but patients whose disorders were not too severe would be sent home or self-isolated and many people were afraid to go to the hospital. This has an impact on health insurance claims which tend to be low during the pandemic and increase significantly after the COVID-19 pandemic in 2023 because people no longer afraid to come to the hospital and some people experiencing diseases due to COVID-19. There are private insurance companies that have experienced losses to the point of product closure, it is necessary to analyze the current severity pattern so that the insurance industry improves again with a more detailed analysis such as analyzing down to each city and analyzing the ages that cause high claims. Collaboration with Social Health Insurance Administration Body can also be one solution to reduce claim severity, the implementation of this cooperation certainly requires a strategy, but the severity pattern and relationship are not necessarily the same as the initial conditions. The Generalized Additive Model (GAM) model will be formed for conditions before and after considering Social Health Insurance Administration Body therfore the implementation of efficiency is on target. The pattern will be analyzed using the GAM method to analyze the relationship between claim severity and the continuous variable age, as well as the categorical variable room class, and province and city in the Java region, Social Health Insurance Administration Body therfore splitbill, and type of claim reimbursement. The model is formed by performing a natural logarithmic transformation on the initial and final severity responses and for the continuous predictor age, smoothing will be carried out using the penalized spline regression method with a division of 24 partitions. There are two models that will be analyzed in the thesis based on the results of the initial GAM model. In the first model, city mapping will be carried out to minimize cities that are less significant for the model. In the second model, city mapping is carried out like GAM Model 1 and adjustments are also made to the room class. The analysis is carried out by comparing each variable with a level basis, C Class I, DKI Jakarta, South Jakarta, insurance only, and cashless. In this thesis, the success of the GAM model in explaining variability and reducing deviations from the basic model ranges from 64% in initial severity and 62% in final severity. DKI Jakarta Province is the highest cause of severity and Central Java is a very ideal province with very low severity and the most effective severity reduction. Big cities such as Semarang and Surabaya, and Sidoarjo Regency need attention for their high severity, while Kudus and Sumedang Regencies have good efficiency. VIP class tends to be less efficient while Class I tends to be more effective than other room classes. Severity efficiency is seen in college age to early working age and retirement age (66.8-70 years).