Key points are not available for this paper at this time.
The convergence of artificial intelligence (AI) and healthcare is reshaping clinical practice, yet this transformation raises pressing questions about scientific rigor and ethical responsibility. This review provides a critical appraisal of research integrity and data ethics considerations specific to AI implementation in integrated healthcare settings. We analyzed peer-reviewed literature from 2019 to 2025, focusing on algorithmic transparency, model validation and reproducibility, bias detection, privacy protection, informed consent paradigms, and governance frameworks. Our analysis reveals a fundamental tension: the data-intensive nature of AI development often conflicts with established principles of patient autonomy and data protection. The opacity of deep learning models challenges conventional standards of scientific transparency, while datasets reflecting historical healthcare disparities risk encoding and amplifying bias. We propose an integrated governance model that aligns technical validation with ethical oversight, emphasizing the need for prospective clinical trials, diverse stakeholder engagement, and adaptive regulatory approaches. This review offers practical guidance for researchers, clinicians, and policymakers navigating the complex intersection of AI innovation and healthcare ethics.
Zeng et al. (Thu,) studied this question.