Continuous Accessibility Validation for Healthcare Web Applications Using AI-Driven CI/CD Integration

Healthcare web applications such as patient portals, telehealth systems, and medication management platforms increasingly rely on continuous deployment practices, yet accessibility validation remains largely dependent on rule-based testing tools with limited detection capability. This paper presents HEALTHA11Y, an AI-driven accessibility validation framework designed for continuous integration and continuous deployment (CI/CD) environments in healthcare software systems. The framework combines a fine-tuned DistilBERT semantic analyzer, a convolutional neural network for visual contrast inspection, and a graph neural network for DOM structure evaluation to identify WCAG 2.1 accessibility violations in healthcare web interfaces. To support evaluation, a domain-specific dataset containing 3,420 annotated healthcare web pages was constructed across patient portals, telehealth interfaces, and pharmacy systems. Experimental results demonstrate that HEALTHA11Y achieves a macro-averaged F1-score of 0.863 and ROC-AUC of 0.924, significantly outperforming traditional rule-based accessibility testing tools. The framework maintains an average CI/CD validation latency of 4.3 seconds per page, enabling practical deployment as an automated accessibility gate within modern healthcare software pipelines. The proposed approach provides a scalable mechanism for continuous accessibility governance, improving the reliability of accessibility compliance verification in regulated digital healthcare environments.