Markov model with machine learning integration for fraud detection in health insurance
This addresses fraud detection for health insurance providers in India, offering a specific incremental improvement over existing methods.
The paper tackled fraud detection in health insurance by developing an improved Markov model integrated with gradient boosting, achieving 97.10% accuracy and an F1-score of 0.8546 on a dataset of 382,587 claims.
Fraud has led to a huge addition of expenses in health insurance sector in India. The work is aimed to provide methods applied to health insurance fraud detection. The work presents two approaches - a markov model and an improved markov model using gradient boosting method in health insurance claims. The dataset 382,587 claims of which 38,082 claims are fraudulent. The markov based model gave the accuracy of 94.07% with F1-score at 0.6683. However, the improved markov model performed much better in comparison with the accuracy of 97.10% and F1-score of 0.8546. It was observed that the improved markov model gave much lower false positives compared to markov model.