Integrating Machine Learning and Visualization for Detection of Healthcare Insurance Fraud

Healthcare fraud is a complex issue, causing billions in losses annually, undermining trust, and complicating care delivery across the system. This study develops a robust fraud detection system that integrates advanced machine learning algorithms with real-time data visualization tools, specifically targeting the healthcare insurance sector. The research involves comprehensive data acquisition, preprocessing, and exploratory data analysis, followed by the application of various machine learning models, including Logistic Regression, AdaBoost, and XGBoost. Among the models evaluated, Logistic Regression demonstrated the highest accuracy of 91.87 percent with a specificity of 99.08 percent, though sensitivity was lower at 24.76 percent. Additionally, the study highlights the importance of real-time monitoring through Power BI dashboards, enhancing decision-making capabilities. By addressing limitations such as class imbalance and focusing on continuous learning, the proposed system aims to provide innovative solutions for proactive fraud detection and improved operational efficiency in healthcare insurance.