Identifying Suspicious Activities in Medical Insurance Claims Using Machine Learning

Abstract

The research methodology involves collecting and preprocessing a comprehensive dataset comprising healthcare claims and associated fraud labels. Multiple machine learning algorithms, including logistic regression, decision trees, random forests, support vector machines, and neural networks, are implemented and evaluated. Performance metrics such as accuracy, precision, recall, and F1 score are used to assess the effectiveness of each model.

The results of the study demonstrate that machine learning techniques exhibit considerable potential in healthcare fraud detection. The comparative analysis reveals variations in performance across different algorithms, highlighting the importance of selecting the appropriate model for specific fraud detection tasks. Moreover, feature engineering and selection techniques are explored to enhance the performance of the models.

In conclusion, this comparative study underscores the effectiveness of machine learning in healthcare fraud detection. It emphasizes the need for tailored approaches and careful consideration of algorithm selection and feature engineering. The best accuracy was achieved as 96.52% with XG Boost algorithm using hyper tuning of various parameters. By leveraging the power of machine learning, healthcare organizations can strengthen their fraud detection capabilities, leading to significant cost savings, enhanced patient care, and improved overall healthcare outcomes.