Prediction of Kidney Failure and Cancer Insurance Claims with Bayesian MCMC

General Background: Health insurance plays a crucial role in mitigating the financial risks of catastrophic illnesses. Specific Background: In Indonesia, chronic kidney disease (CKD) and cancer contribute significantly to the burden of BPJS Health claims, with rising costs reported in recent years. Knowledge Gap: Existing claim estimation models often fail to capture the uncertainty and variability inherent in real-world data. Aims: This study aims to develop a Bayesian model with a Markov Chain Monte Carlo (MCMC) approach to accurately estimate insurance claims for CKD and cancer. Results: Using 2021–2024 data from RSUP Dr. Soeradji Tirtonegoro Klaten, the model successfully estimated outpatient CKD claims at 1649.29 (SD = 19.82) and outpatient cancer claims at 147.68 (SD = 10.18). All model diagnostics indicate strong convergence and accuracy (R-hat = 1.0, ESS > 5000). Novelty: This research applies MCMC-based Bayesian inference with various prior settings (informative to non-informative) and demonstrates robust posterior prediction under different assumptions. Implications: The model provides a credible framework for insurance risk management, improving claim prediction and fiscal planning for health providers and insurers, particularly in managing high-cost diseases within the national health system.Highlight : The Bayesian MCMC model produced accurate and stable estimates of kidney failure and cancer claims (R-hat = 1.0, ESS > 5000). Sensitivity analysis showed the results remained stable despite different priors, indicating a robust model. The best prediction in outpatient CKD (MAPE 5.15%), but less accurate in outpatient cancer (MAPE 50.26%). Keywords : Bayes, Claims, MCMC, PyMC, Python