A Hybrid Model for Detecting Insurance Fraud Using K Means and Support Vector Machine Algorithms
Loading...
Date
2024-10
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Kenyatta University
Abstract
Medical insurance fraud is a significant issue in the healthcare sector, commonly characterized by fraud patterns such as misrepresentation of services, false claims, and identity theft. These patterns contribute to severe data class imbalances, with legitimate claims vastly outnumbering fraudulent ones, complicating effective detection. Current fraud detection methods struggle to address these evolving patterns and manage imbalanced datasets. This study employs a mixed-methods approach, integrating an extensive literature review with quantitative analysis of historical medical claims data. The research develops and evaluates four machine learning models: a standalone Support Vector Machine (SVM), a tuned SVM, a hybrid model combining K-Means clustering with SVM, and a tuned hybrid model. The models were compared using key metrics, including accuracy, precision, recall, and F1 score. Results show that the tuned hybrid model achieved the highest performance with an accuracy of 97.49%, demonstrating its superior ability to detect fraudulent claims compared to the standalone and default hybrid models. Future work will focus on further improving the computational efficiency of the hybrid model and exploring its adaptability to new and evolving fraud patterns in real-time environments. This research significantly advances fraud detection by offering a robust solution that tackles class imbalances and adapts to evolving fraud schemes
Description
A research project submitted in partial fulfilment of the requirements for the award of the degree of Master of Science (Computer Science) in the School of Pure and Applied Sciences of Kenyatta University.
Supervisor
Abraham Matheka Mutua