An Explainable Generative AI Framework for Insurance Claims Intelligence and Risk Assessment
DOI:
https://doi.org/10.63282/3050-922X.IJERET-V7I3P103Keywords:
Insurance Claims Intelligence, Insurance Risk Assessment, Explainable Artificial Intelligence (XAI), Generative AI and Fraud DetectionAbstract
Insurance claims are becoming more complex, necessitating more sophisticated, clear, and automated risk evaluation processes. This research proposes a framework for the analysis of insurance claims and human decision support that combines the use of ML, explainable AI (XAI), and generative AI. The Insurance Claims and Policy Data dataset has 53,503 records, and these have been preprocessed using data cleaning, feature engineering, label encoding, and feature scaling of the data in a public repository. Several baseline models, namely Extra Trees Classifier (ETC), Gradient Boosting Classifier (GBC) and Multi-Layer Perceptron (MLP), are built and fused using a Hybrid Soft Voting Ensemble. To assess the proposed framework, the following metrics are used: Accuracy, Balanced Accuracy, Precision, Recall, Weighted F1-Score, Cohen's Kappa, MCC and ROC-AUC. The experimental findings show that compared to the individual models, the hybrid ensemble model has the best classification accuracy of 100%. Moreover, SHAP-based explainability helped to reveal the most critical features to capturing insurance risk prediction and Generative AI facilitated what-if risk scenario simulation, real-time claims intelligence, settlement estimation and automated decision support. The envisioned framework offers accurate, robust, and interpretable solution for smart insurance claims handling and risk assessment.
References
[1] A. R. Aitha, “Explainable Agentic AI Framework for Automated Insurance Fraud Detection and Predictive Risk Intelligence,” Int. J. Adv. Res. Comput. Sci. Technol., vol. 9, no. 03, May 2026, doi: 10.15662/IJARCST.2026.0903009.
[2] E. Brati, A. Braimllari, and A. Gjeçi, “Machine Learning Applications for Predicting High-Cost Claims Using Insurance Data,” Data, vol. 10, no. 6, Jun. 2025, doi: 10.3390/data10060090.
[3] K. Amistapuram and P. Pandey, “Generative AI for Claims Exceptions and Investigations: Enhancing Resolution Efficiency in Complex Insurance Processes,” Lex localis - J. Local Self-Government, vol. 23, no. S6, pp. 7320–7337, Oct. 2025, doi: 10.52152/y5j9v602.
[4] K. McDonnell, B. Sheehan, and F. Murphy, “Bridging transparency in insurance claims prediction: A comparative study of explainable AI and traditional linear models using vehicle telematics data,” Technol. Forecast. Soc. Change, vol. 223, p. 124418, Feb. 2026, doi: 10.1016/j.techfore.2025.124418.
[5] V. Rengaraj, Y. Lnu, Y. Patel, and S. Vinod, “Decoding-Enhanced Bidirectional Encoder Representations with Attention-Based Pooling Mechanism for Insurance Fraud Detections,” in 2026 3rd International Conference on Research Methodologies in Knowledge Management, Artificial Intelligence and Telecommunication Engineering (RMKMATE), 2026, pp. 1–9. doi: 10.1109/RMKMATE69073.2026.11518574.
[6] P. A. Kumar and S. Sountharrajan, “Insurance claims estimation and fraud detection with optimized deep learning techniques,” Sci. Rep., vol. 15, no. 1, p. 27296, Jul. 2025, doi: 10.1038/s41598-025-12848-0.
[7] J. Mishra, S. Rai, B. Moharana, V. K. Singh, S. Dey, and T. Sarkar, “Revolutionizing Financial Fraud Detection in Healthcare Insurance Through Blockchain & AI-Driven Predictive Analysis,” in 2025 International Conference on Technology Enabled Economic Changes (InTech), IEEE, Feb. 2025, pp. 1516–1522. doi: 10.1109/InTech64186.2025.11198190.
[8] M. G. Rao, V. B. S, P. Rao H, M. Mascarenhas, A. P. Kumar, and S. Shanbhag, “A Data Fusion–Driven Framework for Predicting Insurance Risk Using Explainable Ensemble and Deep Learning Approaches,” in 2026 International Conference on Artificial Intelligence and Data Engineering (AIDE), IEEE, Feb. 2026, pp. 851–856. doi: 10.1109/AIDE69088.2026.11544448.
[9] L. Suraiya, B. P. C, F. R. Sumarna, J. R. Machireddy, A. A, and M. C. Janaki, “Machine Learning-Based Risk Prediction Framework for Insurance Fraud Detection and Claim Optimization,” in 2025 International Conference on Recent Innovation in Science Engineering and Technology (ICRISET), IEEE, Aug. 2025, pp. 1–6. doi: 10.1109/ICRISET64803.2025.11254689.
[10] L. P. P and G. Ramkumar, “Insurance Claim Prediction and Risk Assessment Using Machine Learning,” in 2025 1st International Conference on Emerging Innovation and Digital Technology (ICEIDT), IEEE, Nov. 2025, pp. 288–292. doi: 10.1109/ICEIDT66693.2025.11473722.
[11] B. Al-Attar et al., “Supervised Machine Learning Framework for Detection, Classification, and Risk Assessment of Fraudulent Legal Claims in Insurance Litigation,” in 2025 3rd International Conference on Cyber Resilience (ICCR), IEEE, Jul. 2025, pp. 1–7. doi: 10.1109/ICCR67387.2025.11291936.
[12] A. Khan, K. Dhungana, S. Gyawali, and I. Sharma, “Forecasting Automobile Insurance Claims: Examining the Interplay of Policyholder Traits for Enhanced Predictive Insights,” in 2024 2nd International Conference on Intelligent Data Communication Technologies and Internet of Things (IDCIoT), IEEE, Jan. 2024, pp. 881–886. doi: 10.1109/IDCIoT59759.2024.10467373.
[13] K. P. Rijanto, D. A. Putri, P. Calvin, I. S. Edbert, and D. Suhartono, “Machine Learning Algorithms with Oversampling Methods for Automobile Insurance Fraud Detection,” in 2023 IEEE 21st Student Conference on Research and Development (SCOReD), IEEE, Dec. 2023, pp. 138–143. doi: 10.1109/SCOReD60679.2023.10563582.
[14] S. Raval, “Insurance Claims and Policy Data,” Kaggle. [Online]. Available: https://www.kaggle.com/datasets/ravalsmit/insurance-claims-and-policy-data
[15] A. Polamarasetti et al., “Enhancing Risk Assessment in Auto Insurance with Data-Driven Insights using Machine Learning,” J. Contemp. Educ. Theory Artif. Intell., vol. 3, 2023, doi: 10.71010/2023/JCETAI-e104-1.
[16] S. K. Majhi, S. Bhatachharya, R. Pradhan, and S. Biswal, “Fuzzy clustering using salp swarm algorithm for automobile insurance fraud detection,” J. Intell. Fuzzy Syst., vol. 36, no. 3, pp. 2333–2344, 2019.
[17] H. Moon, Y. Pu, C. Ceglia, and others, “A predictive modeling for detecting fraudulent automobile insurance claims,” Theor. Econ. Lett., vol. 9, no. 06, p. 1886, 2019.
[18] A. S. Elbhrawy, M. A. Belal, and M. S. Hassanein, “AIRA-ML: Auto Insurance Risk Assessment-Machine Learning Model using Resampling Methods,” Int. J. Adv. Comput. Sci. Appl., vol. 14, no. 9, pp. 633–641, 2023, doi: 10.14569/IJACSA.2023.0140966.
