Tara Templin, Sean Sylvia, and collaborators have published a comprehensive scoping review in PLOS Digital Health examining six key challenges in generative AI for healthcare: bias, privacy, hallucination, adversarial attacks, regulatory compliance, and model interpretability.
The research, supported by the National Science Foundation and Gillings Innovation Labs, surveyed 120 articles to identify solutions and best practices for digital health practitioners implementing AI technologies.
As generative AI systems become more prevalent in healthcare—from clinical documentation to patient communication to diagnostic support—understanding and addressing these challenges is essential. The paper provides actionable guidance for practitioners seeking to deploy these technologies responsibly.