WU Chenxi, FANG Qiquan, ZHANG Ying

DOI: https://doi.org/10.5281/zenodo.20685682

To address medical insurance fraud, this study developed a Federated Generative Adversarial Network (FGAN) model, integrating federated learning with GANs. Using the Shenzhen Cup inpatient reimbursement dataset, we performed standardized preprocessing and evaluated the model against logistic regression, random forest, a standalone GAN, and a baseline federated learning method. Results indicate that the FGAN model achieved balanced performance, with precision, recall, F1-score, and AUC reaching 0.927, 0.892, 0.909 and 0.986 respectively. Model performance also improved consistently with more training epochs. By enabling collaborative modeling across institutions without sharing raw data, FGAN mitigates sample limitations inherent in isolated federated learning setups. This work demonstrates the effectiveness and practical value of FGAN for detecting and preventing medical insurance fraud.

Federated Learning, Generative Adversarial Networks, Logistic Regression, Random Forest, Healthcare Insurance, Anti-Fraud

By combining distributed sample generation with cross-agency collaboration, the Federated Generation Adversarial Network (FGAN) has successfully resolved data isolation, sample acquisition, and privacy issues in health insurance anti-fraud. The anti-fraud system achieved an accuracy rate of up to 95.5% and successfully identified cross-agency collaborative fraud cases. With the actual use of FGAN, prediction performance improves, and FGAN achieves expected accuracy with very little training, greatly reducing costs. Implementation of FGAN is expected to save millions in health insurance funds, verifying its technological feasibility and social value.

[1] T. Awosika, R. M. Shukla and B. Pranggono, "Transparency and Privacy: The Role of Explainable AI and Federated Learning in Financial Fraud Detection," in IEEE Access, vol. 12, pp. 64551-64560, 2024. https://ieeexplore.ieee.org/abstract/document/10509682
[2] Brisimi T S, Chen R, Mela T, et al. Federated learning of predictive models from federated electronic health records[J]. International journal of medical informatics, 2018, 112: 59-67. https://doi.org/10.1016/j.ijmedinf.2018.01.007
[3] Abdul Salam, M., Fouad, K.M., Elbably, D.L.et al. Federated learning model for credit card fraud detection with data balancing techniques. Neural Comput & Applic 36, 6231–6256 (2024). https://link.springer.com/article/10.1007/s00521-023-09410-2
[4] W. Du, H. Wang, J. Shen, G. Meng, H. Li and W. Zhou, "Insurance Anti-fraud based on FL-WOE Encoding for Vertical Federated Learning," 2024 IEEE International Conference on Big Data (BigData), Washington, DC, USA, 2024, pp. 7717-7724. https://link.springer.com/article/10.1007/s11433-019-1528-y
[5] Y. Tang and Z. Liu, "A Credit Card Fraud Detection Algorithm Based on SDT and Federated Learning," in IEEE Access, vol. 12, pp. 182547-182560, 2024. https://ieeexplore.ieee.org/document/10742353
[6] Ugo Fiore, Alfredo De Santis, Francesca Perla, Paolo Zanetti, Francesco Palmieri，Using generative adversarial networks for improving classification effectiveness in credit card fraud detection, Information Sciences, Volume 479,2019,Pages 448-455,ISSN 0020-0255. https://api.semanticscholar.org/CorpusID:59528925
[7] XIAO Yi , XIAO Jin , LIU John , WANG Shouyang. A MULTISCALE MODELING APPROACH INCORPORATING ARIMA AND ANNS FOR FINANCIAL MARKET VOLATILITY FORECASTING. Journal of Systems Science and Complexity, 2014, 27(1): 225-236. https://doi.org/10.1007/s11424-014-3305-4
[8] Li T, Sahu A K, Talwalkar A, et al. Federated learning: Challenges, methods, and future directions[J]. IEEE Signal Processing Magazine, 2020, 37(3): 50-60. https://ieeexplore.ieee.org/document/9084352
[9] Sheller M J, Edwards B, Reina G A, et al. Federated learning in medicine: facilitating multi-institutional collaborations without sharing patient data[J]. Scientific reports, 2020, 10(1): 1-12. https://doi.org/10.1038/s41598-020-69250-1
[10] Chawla N V, Bowyer K W, Hall L O, et al. SMOTE: synthetic minority over-sampling technique[J]. Journal of artificial intelligence research, 2002, 16: 321-357. https://api.semanticscholar.org/CorpusID:1554582
[11] 支宇晗,胡强,谢肖飞,等. 联邦学习中基于数据选择的模型修复方法研究[J/OL].中国科学:信息科学, 1-26[2026-05-26]. https://link.cnki.net/urlid/11.5846.TP.20260522.1104.004
[12] 杜倩, 刘鸿宇, 胡琦. 社会医疗保险基金欺诈行为的扎根理论研究——基于58个医保欺诈刑事案件分析[J]. 法制与经济, 2020(8): 70-71+75.
[13] 菅银龙,陈学斌,景忠瑞,等.基于条件生成对抗网络的联邦学习中数据增强方案[J/OL].计算机应用,1-16[2025-05-16].
[14] 陈志强.基于GAN模型的电网用电数据安全生成研究[D].华北电力大学(北京),2023.
[15] 余锋,林庆新,林晖,等.基于生成对抗网络的隐私增强联邦学习方案[J].网络与信息安全学报,2023,9(03):113-122.
[16] 瞿祥谋.基于生成对抗网络的联邦学习非独立同分布数据问题研究[D].重庆大学,2022.
[17] 葛广为. 联邦学习在医疗数据隐私保护中的应用研究[J].信息记录材料,2026,27 (10):181-183.DOI:10.16009/j.issn.1009-5624.2026.10.058.
[18] 蒋大锐,吕峻闽,徐胜超. 基于生成式对抗网络的网络安全态势感知[J/OL].计算机测量与控制, 1-10[2026-05-26]. https://link.cnki.net/urlid/11.4762.tp.20260122.1740.002
[19] 陈石轩,王博琛,赵志林. 基于GAN的恶意PE文件对抗样本生成模型研究[J].电脑与信息技术,2025,33 (6):69-72+127.DOI:10.19414/j.cnki.1005-1228.2025.06.004.
[20] 考兴楷,苗莉,郑远硕. 基于联邦学习的安全边缘缓存优化机制[J/OL].计算机系统应用, 1-16[2026-05-26]. https://doi.org/10.15888/j.cnki.csa.010197

WU Chenxi, FANG Qiquan, ZHANG Ying (2026). The Application of Federated Generative Adversarial Networks in Medical Insurance Anti-Fraud. International Journal of Computer Science Engineering Techniques, 10(3), 149–157. ISSN: 2455-135X. DOI: https://doi.org/10.5281/zenodo.20685682