Open Access

ARTICLE

Cyber-Integrated Predictive Framework for Gynecological Cancer Detection: Leveraging Machine Learning on Numerical Data amidst Cyber-Physical Attack Resilience

Muhammad Izhar^1,*, Khadija Parwez², Saman Iftikhar³, Adeel Ahmad⁴, Shaikhan Bawazeer³, Saima Abdullah⁴

1 Department of Computer Science and Information Technology, Superior University, Lahore, 54000, Pakistan
2 Department of Computing and Technology, IQRA University, Sector H-9, Islamabad, 04436, Pakistan
3 Faculty of Computer Studies, Arab Open University, Riyadh, 84901, Saudi Arabia
4 Department of Computer Science, Faculty of Computing, The Islamia University of Bahawalpur, Bahawalpur, 63100, Pakistan

* Corresponding Author: Muhammad Izhar. Email:

Journal on Artificial Intelligence 2025, 7, 55-83. https://doi.org/10.32604/jai.2025.062479

Received 19 December 2024; Accepted 31 March 2025; Issue published 25 April 2025

Abstract

The growing intersection of gynecological cancer diagnosis and cybersecurity vulnerabilities in healthcare necessitates integrated solutions that address both diagnostic accuracy and data protection. With increasing reliance on IoT-enabled medical devices, digital twins, and interconnected healthcare systems, the risk of cyber-physical attacks has escalated significantly. Traditional approaches to machine learning (ML)–based diagnosis often lack real-time threat adaptability and privacy preservation, while cybersecurity frameworks fall short in maintaining clinical relevance. This study introduces HealthSecureNet, a novel Cyber-Integrated Predictive Framework designed to detect gynecological cancer and mitigate cybersecurity threats in real time simultaneously. The proposed model employs a three-tier ML architecture incorporating Gradient Boosting and Support Vector Machines (SVMs) for accurate cancer classification, combined with an adaptive anomaly detection layer leveraging Mahalanobis Distance and severity scoring for threat prioritization. To enhance resilience, the framework integrates Zero Trust principles and Federated Learning (FL), enabling secure, decentralized model training while preserving patient privacy and meeting compliance with HIPAA (Health Insurance Portability and Accountability Act) and GDPR (General Data Protection Regulations). Experimental evaluation using a real-world healthcare cybersecurity dataset demonstrated high accuracy (95.2%), precision (94.3%), recall (91.7%), and AUC-ROC (Area Under the Curve-Receiver Operating Characteristic) (0.94), with a low false positive rate (3.6%). HealthSecureNet outperforms traditional models such as SVM, Random Forest (RF), and k-NN (k-Nearest Neighbor) in both anomaly detection and severity classification accuracy. Its adaptive thresholding and response prioritization mechanisms make it suitable for dynamic healthcare environments, enabling early cancer detection and proactive cyber threat mitigation without compromising performance or regulatory standards. This research contributes a robust, dual-purpose solution that enhances both clinical diagnostics and cybersecurity, setting a precedent for future AI (Artificial Intelligence)-driven healthcare systems.

---

Keywords

---