Journal of Preventive Medicine and Public Health

Original Articles

Development of Machine Learning Models to Predict Health Insurance Claim Costs Among Older Indonesians: A Retrospective Predictive Modeling Study: Yeni Mahwati, Dhihram Tenrisau, Syarif Rahman Hasibuan, Bhirau Wilaksono, Yeni Indriyani, Andi Afdal Abdullah, Halik Malik, Andi Alfian Zainuddin; J Prev Med Public Health. 2026;59(2):132-142. Published online January 6, 2026; DOI: https://doi.org/10.3961/jpmph.25.350

737 View
96 Download

Abstract Summary PDF: Objectives
The objective of this study was to develop machine learning models to predict health insurance claim costs among older adults in Indonesia.
Methods
This study utilized secondary data from the Indonesian National Health Insurance program (Jaminan Kesehatan Nasional [JKN]) spanning 2017 to 2023. Three modeling techniques—linear regression, random forest, and XGBoost—were employed to predict individual claim costs. Model performance was assessed using the root mean square error (RMSE), coefficient of determination (R²), and mean absolute error (MAE). Additionally, variable importance analysis was conducted to identify key predictors.
Results
XGBoost with 500 boosting rounds yielded the best performance, with an RMSE of 11 360 283, an R² of 0.81, and an MAE of 4 485 917, outperforming both linear regression (RMSE, 13 710 035; R²=0.72) and random forest (RMSE, 12 434 238; R²=0.78). Notably, outpatient care was identified as the most consistent predictor across all models. Other significant predictors included length of stay (LOS), diagnosis type (International Classification of Diseases, 10th revision chapter), facility type, facility classification, and severity of illness, particularly for moderate cases. Although LOS and diagnosis type were important predictors, these findings should be interpreted in the context of Indonesia’s fixed Indonesian Case-Based Groups payment system.
Conclusions
XGBoost provides reliable predictions of claim costs among older adults, capturing clinical, utilization, and structural drivers. These findings can inform targeted interventions, improve chronic disease management, optimize the referral system, and support integration of predictive tools into JKN to enhance responsiveness and promote sustainable, equitable financing.; Summary

Key Message
Machine learning models, particularly XGBoost, demonstrated superior performance in predicting healthcare costs among older adults in Indonesia. Nonlinear relationships between outpatient visits, severity, and diagnoses highlight the limitations of conventional linear approaches. These findings support the integration of advanced predictive models into national health insurance systems to improve cost management and resource allocation.

Operational Definitions of Colorectal Cancer in the Korean National Health Insurance Database: Hyeree Park, Yu Rim Kim, Yerin Pyun, Hyundeok Joo, Aesun Shin; J Prev Med Public Health. 2023;56(4):312-318. Published online May 31, 2023; DOI: https://doi.org/10.3961/jpmph.23.033

10,060 View
197 Download
7 Web of Science
7 Crossref

Abstract Summary PDF Supplementary Material

Objectives
We reviewed the operational definitions of colorectal cancer (CRC) from studies using the Korean National Health Insurance Service (NHIS) and compared CRC incidence derived from the commonly used operational definitions in the literature with the statistics reported by the Korea Central Cancer Registry (KCCR).
Methods
We searched the MEDLINE and KoreaMed databases to identify studies containing operational definitions of CRC, published until January 15, 2021. All pertinent data concerning the study period, the utilized database, and the outcome variable were extracted. Within the NHIS-National Sample Cohort, age-standardized incidence rates (ASRs) of CRC were calculated for each operational definition found in the literature between 2005 and 2019. These rates were then compared with ASRs from the KCCR.
Results
From the 62 eligible studies, 9 operational definitions for CRC were identified. The most commonly used operational definition was “C18-C20” (n=20), followed by “C18-C20 with claim code for treatment” (n=3) and “C18-C20 with V193 (code for registered cancer patients’ payment deduction)” (n=3). The ASRs reported using these operational definitions were lower than the ASRs from KCCR, except for “C18-C20 used as the main diagnosis.” The smallest difference in ASRs was observed for “C18-C20,” followed by “C18- C20 with V193,” and “C18-C20 with claim code for hospitalization or code for treatment.”
Conclusions
In defining CRC patients utilizing the NHIS database, the ASR derived through the operational definition of “C18-C20 as the main diagnosis” was comparable to the ASR from the KCCR. Depending on the study hypothesis, operational definitions using treatment codes may be utilized.

Summary

Korean summary
- 국민건강보험공단 자료 기반 연구에서 사용된 대장암의 조작적 정의를 검토하고 각 조작적 정의에 따른 대장암의 연령표준화발생률을 중앙암등록본부에서 보고한 수치와 비교하였다. - 62개의 출판된 문헌을 검토하여 9개의 조작적 정의를 확인하였고 이 중 "C18-C20"이 가장 일반적으로 사용되었다. - "C18-C20"을 주진단으로 정의한 경우의 연령표준화발생률은 중앙암등록본부에서 보고한 연령표준화발생률과 가장 유사하였다.

Citations

Citations to this article as recorded by

RA and risk of colorectal cancer: a nationwide cohort study
Keun Hye Jeon, Jinhyung Jung, Mi Hee Cho, Dagyeong Lee, Kyungdo Han, In Young Cho, Dong Wook Shin
Rheumatology.2026;[Epub] CrossRef
Who Dies Alone? Demographics, Underlying Diseases, and Healthcare Utilization Patterns of Lonely Death Individuals in Korea
Haibin Bai, Jae-ryun Lee, Min Jung Kang, Young-Ho Jun, Hye Yeon Koo, Jieun Yun, Jee Hoon Sohn, Jin Yong Lee, Hyejin Lee
Journal of Preventive Medicine and Public Health.2025; 58(2): 218. CrossRef
Validation of Diagnostic Code Definitions for Identifying Patient in Korean Health Insurance Claims Data: A Scoping Review
Kyoung-Hoon Kim
Journal of Korean Medical Science.2025;[Epub] CrossRef
Worsening of health disparities across COVID-19 pandemic stages in Korea
Hyejin Lee, Hyunwoo Nam, Jae-ryun Lee, Hyemin Jung, Jin Yong Lee
Epidemiology and Health.2024; 46: e2024038. CrossRef
An epidemic of cataract surgery in Korea: the effects of private health insurance on the National Health Insurance Service
Hyejin Lee, Soo-Hee Hwang, Choon-Seon Park, Seol-Hee Chung, Catherine L. Chen, Jin Yong Lee, Jin Soo Lee
Epidemiology and Health.2024; 46: e2024015. CrossRef
National Trends in Rotavirus Enteritis among Infants in South Korea, 2010–2021: A Nationwide Cohort
Hyun Jee Lee, Yujin Choi, Jaeyu Park, Yong-Sung Choi, Dong Keon Yon, Do Hyun Kim
Children.2023; 10(9): 1436. CrossRef
Real-world data analysis on effectiveness of integrative therapies: A practical guide to study design and data analysis using healthcare databases
Ye-Seul Lee, Yoon Jae Lee, In-Hyuk Ha
Integrative Medicine Research.2023; 12(4): 101000. CrossRef

English Abstracts

The Determinants of Purchasing Private Health Insurance in Korean Cancer Patients.: Jin Hwa Lim, Sung Gyeong Kim, Eun Mi Lee, Sin Young Bae, Jae Hyun Park, Kui Son Choi, Myung Il Hahm, Eun Cheol Park; J Prev Med Public Health. 2007;40(2):150-154.; DOI: https://doi.org/10.3961/jpmph.2007.40.2.150

7,695 View
67 Download
13 Crossref

Abstract PDF

OBJECTIVES
The aim of this study is to identify factors determining the purchase of private health insurance under the mandatory National Health Insurance(NHI) system in Korea. METHODS: The data were collected by the National Cancer Center in Korea. It includes cancer patients who were newly diagnosed with stomach (ICD code, C16), lung(C33-C34), liver (C22), colorectal cancer(C18-C20) or breast(C50) cancer. Data were gathered from the hospital Order Communication System (OCS), medical records, and face-to-face interviews, using a structured questionnaire. Clinical, socio-demographic and private health insurance related factors were also gathered. RESULTS: Overall, 43.9% of patients had purchased one or more private health insurance schemes related to cancer, with an average monthly premium of won 65,311 and an average benefit amount of won 19 million. Females, younger aged, high income earners, national health insurers and metropolitan citizens were more likely to purchase private health insurance than their counterparts. CONCLUSIONS: About half of Korean people have supplementary private health insurance and their benefits are sufficient to cover the out-of-pocket fees required for cancer treatment, but inequality remains in the purchase of private health insurance. Further studies are needed to investigate the impacts of private health insurance on NHI, and the relationship between cancer patients' burden and benefits.

Summary

Citations

Citations to this article as recorded by

Out-of-pocket costs in gastrointestinal cancer patients: Lack of a perfectly framed problem contributing to financial toxicity
Roberto Bordonaro, Dario Piazza, Concetta Sergi, Stefano Cordio, Salvatore Tomaselli, Vittorio Gebbia
Critical Reviews in Oncology/Hematology.2021; 167: 103501. CrossRef
Assessing determinants of health care prepayment in China: Economic growth or government willingness? New evidence from the continuous wavelet analysis
Ying Zhang, Rui Wang, Xinyi Yao
The International Journal of Health Planning and Management.2019;[Epub] CrossRef
Evaluating financial performance of insurance companies using rating transition matrices
Abhijit Sharma, Diara Md Jadi, Damian Ward
The Journal of Economic Asymmetries.2018; 18: e00102. CrossRef
Having Private Cancer Insurance in Korea: Gender Differences
Ki-Bong Yoo, Jin-Won Noh, Young Dae Kwon, Kyoung Hee Cho, Young Choi, Jae-Hyun Kim
Asian Pacific Journal of Cancer Prevention.2015; 16(17): 7981. CrossRef
Equity in health care: current situation in South Korea
Hong-Jun Cho
Journal of the Korean Medical Association.2013; 56(3): 184. CrossRef
Factors Associated with the Middle-aged or the Old-aged Koreans' Enrollment in Private Health Insurance
Hyo-Jin Kim, Jae-Hee Lee
The Journal of the Korea Contents Association.2012; 12(12): 683. CrossRef
Comparison of Health Promotion Behavior in Middle aged Rural Residents by Cancer Screening Participation
Myung Suk Lee
Journal of Korean Academy of Community Health Nursing.2010; 21(1): 43. CrossRef
The Effect of Catastrophic Health Expenditure on the Transition to Poverty and the Persistence of Poverty in South Korea
Eun-Cheol Song, Young-Jeon Shin
Journal of Preventive Medicine and Public Health.2010; 43(5): 423. CrossRef
The Effect of Health Care Expenditure on Income Inequality
Eun-Cheol Song, Chang-Yup Kim, Young-Jeon Shin
Korean Journal of Health Policy and Administration.2010; 20(3): 36. CrossRef
Impact of supplementary private health insurance on stomach cancer care in Korea: a cross-sectional study
Dong Wook Shin, Kee-Taig Jung, Sung Kim, Jae-Moon Bae, Young-Woo Kim, Keun Won Ryu, Jun Ho Lee, Jae-Hyung Noh, Tae-Sung Sohn, Young Ho Yun
BMC Health Services Research.2009;[Epub] CrossRef
Comparison of Cancer Survival by Age Group for 1997 and for 2002: Application of Period Analysis using the National Cancer Incidence Database
Seon-Hee Yim, Kyu-Won Jung, Young-Joo Won, Hyun-Joo Kong, Hai-Rim Shin
Journal of Preventive Medicine and Public Health.2008; 41(1): 17. CrossRef
Strengthening Causal Inference in Studies using Non-experimental Data: An Application of Propensity Score and Instrumental Variable Methods
Myoung-Hee Kim, Young Kyung Do
Journal of Preventive Medicine and Public Health.2007; 40(6): 495. CrossRef
Effects of Private Health Insurance on Health Care Utilization and Expenditures in Korean Cancer Patients: Focused on 5 Major Cancers in One Cancer Center
Jin Hwa Lim, Kui Son Choi, Sung Gyeong Kim, Eun-Cheol Park, Jae Hyun Park
Journal of Preventive Medicine and Public Health.2007; 40(4): 329. CrossRef

Effects of Air Pollution on Asthma in Seoul: Comparisons across Subject Characteristic.: Sun Young Kim, Ho Kim, Jaiyong Kim; J Prev Med Public Health. 2006;39(4):309-316.

3,067 View
65 Download

Abstract PDF: OBJECTIVES
Korean epidemiological studies have used reduced samples according to the subject's characteristics, such as the health services provided, the historical note with asthma, and age, to examine the acute effect of air pollution on asthma using the Korean National Health Insurance records. However, there have been few studies on whether the effects shown in these reduced samples are different from those of all samples. This study compared the effects of air pollution on asthma attacks in three reduced samples with those of entire samples. METHODS: The air pollution data for PM10, CO, SO2, NO2, and O3 and weather conditions including temperature, relative humidity, and air pressure in Seoul, 2002, were obtained from outdoor monitoring stations in Seoul. The emergency hospital visits with an asthma attack in Seoul, 2002 were extracted from the Korean National Health Insurance records. From these, the reduced samples were created by health service, historical notes with asthma, and age. A case-crossover design was adopted and the acute effects of air pollution on asthma were estimated after adjusting for weather, time trend, and seasonality. The model was applied to each reduced sample and the entire sample. RESULTS: With respect to the health service, the effects on outpatients were similar to those for the total sample but were different for inpatients. These similar effect sizes were also observed in the reduced samples according to the historical note with asthma and age. The relative risks of PM10, CO, SO2, NO2, and O3 among the reduced and entire samples were 1.03, 1.04-1.05, 1.02-1.03, 1.04-1.06, and 1.10-1.17, respectively. CONCLUSIONS: There was no clear evidence to show a difference between the reduced samples and the entire samples.; Summary

First
Prev
Page of 1
Next
Last

Search