A Comparative Analysis for Identifying the Fraudulent Healthcare Claims

Medical expenses have sharply increased in recent times, primarily due to the rise in the elderly population. To effectively manage resources for all groups of people, health insurance policies have been developed. These programs can help educate citizens about cost control, quality fees, expenses, budgeting, and future payments, enabling them to utilize healthcare services at affordable costs. Supervised machine learning algorithms are playing a dominant role in identifying fraudulent claims submitted by hospitals, patients, or doctors. This is significantly reducing the financial burden on the public, the government, and private health insurance providers. To prevent mismanagement of costs in the healthcare sector, there is a growing demand to identify falsified (fraudulent) claims. A considerable amount of resources is being spent on developing fraud control systems to deal with fraudulent practitioners and planned illegal patterns. In this study, a comparative analysis using machine learning algorithms like Random Forest (RF), Support Vector Machine (SVM), Logistic Regression (LR), and Artificial Neural Networks (ANN) was conducted to analyze behaviors and identify fraudulent and non-fraudulent claims in the Center for Medicaid and Medicare Services (CMS) data. The results show that SVM had superior accuracy compared to other algorithms. The analysis also includes different metrics to evaluate relative performance. Overall, the development and implementation of effective fraud control systems and the use of advanced machine learning algorithms can help improve the management of healthcare costs and ensure that resources are utilized efficiently and effectively