VALIDATING ML MODELS FOR GOVERNMENT HEALTH DATA PRIVACY AGAINST CYBER THREATS USING MCDM TECHNIQUES

Abstract

In the era of digital governance, the protection of sensitive government health data has become a critical priority due to the increasing frequency and sophistication of cyber threats. Machine Learning (ML) models are widely adopted for detecting and preventing such attacks; however, validating the effectiveness of these models requires a structured and multi-perspective evaluation. This research proposes a hybrid Multi-Criteria Decision-Making (MCDM) framework using the Analytic Hierarchy Process (AHP) and Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) to assess and validate ML models based on multiple criteria including accuracy, robustness, interpretability, response time, and resilience against data breaches. The results provide a comprehensive and objective basis for selecting the most suitable ML model to safeguard government health data, enhancing both privacy assurance and cybersecurity readiness. This study not only offers a novel MCDM-based validation framework but also contributes to the development of secure and trustworthy AI systems for public sector health data protection.