PENERAPAN ALGORITMA RANDOM FOREST DALAM PENCARIAN POLA KLAIM BPJS RUMAH SAKIT UMUM KUMALA SIWI MIJEN KUDUS

The persistently high number of pending BPJS Kesehatan claims in hospitals impacts cash flow and financial risk, necessitating data-driven solutions to improve the effectiveness of claims management. This study aims to analyze the factors influencing BPJS claims status and test the performance of the Random Forest algorithm in classifying eligible and pending claims. The results show that feature analysis identified total hospital charges and LOS as the most dominant variables, followed by diagnosis and treatment codes. The Random Forest model, with 30 decision trees, achieved 96.88% accuracy with balanced precision and recall, despite imbalanced data (520 eligible claims and 24 pending). Claim patterns indicate that claims with high costs, long LOS, and complex diagnoses and treatments are at higher risk of delay. This study demonstrates Random Forest's technical superiority and managerial benefits as an early warning system, supporting data-driven decisions, and improving the efficiency of claims management and hospital cash flow.