Sultana, Nasrin and Nasir, Abu and Hasan, Asif (2025) HRIS-Integrated Fraud Detection for U.S. Home Health Agency Workforce Compliance: Identifying Ghost Worker Schemes and Payroll Irregularities in Medicare and Medicaid-Funded Care Programs. International Journal of Computational and Experimental Science and Engineering, 11 (1). ISSN 2149-9144

Text final+5496.pdf Download (777kB)

Abstract

Background: Workforce-related fraud — ghost workers, post-termination payments, duplicate hours, and payroll–service mismatches — poses a distinct compliance risk for U.S. home health agencies operating under Medicare and Medicaid, separate from claims-level billing fraud. Human resource information systems (HRIS), scheduling, electronic visit verification (EVV), and payroll platforms are typically operated independently, making cross-system inconsistencies hard to detect through any single system's audit. Objective: This study proposes and empirically evaluates an HRIS-integrated fraud detection framework that cross-reconciles employee identity, scheduling, EVV, and payroll data to detect ghost-worker schemes and payroll irregularities, comparing rule-based detection against machine learning and network-based collusion indicators. Methods: A reproducible synthetic dataset of 20,264 employee-pay-period records was constructed for 900 employees and 500 patients over 12 months, including 14 simulated collusion rings sharing bank accounts or addresses and 719 fraudulent records (3.55%) across five documented typologies. Rule-based cross-system indicators, Random Forest, XGBoost, and Isolation Forest were comparatively evaluated on a temporally held-out test period, with SHAP explainability applied to the strongest classifier. Results: Rule-based indicators alone achieved only 45.6% recall (F1 = 0.416) despite perfect precision, missing more than half of injected fraud. XGBoost achieved the strongest overall performance (F1 = 0.974, ROC-AUC = 0.9998), detecting 100% of ghost-worker and post-termination-payment cases and above 89% recall on every typology. The payroll–EVV hour gap and maximum inter-visit travel distance were the dominant predictors. Conclusions: Cross-system reconciliation alone is necessary but insufficient; layering machine learning on top of HRIS-EVV-payroll integration substantially improves ghost-worker and payroll-irregularity detection over threshold rules alone, while network-based collusion indicators require care to avoid conflating legitimate shared household identifiers with fraud. Validation on operational agency data is the essential next step.

Item Type:	Article
Uncontrolled Keywords:	Human Resource Information Systems (HRIS), Healthcare fraud detection, ghost workers, Electronic Visit Verification (EVV), Medicare and Medicaid compliance, payroll irregularities
Subjects:	H Social Sciences
Depositing User:	admin eprints
Date Deposited:	20 Oct 2026 08:00
Last Modified:	20 Oct 2026 08:00
URI:	http://eprints.umsida.ac.id/id/eprint/17221

Actions (login required)

View Item