Generative AI in Healthcare: Applications and Challenges
TL;DR
Generative AI demonstrably accelerates diagnostic workflows, augments scarce clinical datasets, personalizes communication, and supports discovery pipelines, and the paper concludes with a translational path and research priorities aimed at closing these gaps.
Abstract
This scoping review synthesizes recent literature on the applications and challenges of generative artificial intelligence (AI) in healthcare, with the aim of providing a structured overview that is useful to clinicians, data scientists, and health-system decision-makers. Using a Preferred Reporting Items for Systematic Reviews and Meta-Analyses-guided search across PubMed, IEEE Xplore, ACM Digital Library, ScienceDirect, Scopus, and Google Scholar, we identified and thematically analyzed 86 studies. The review organizes applications into seven themes—medical image processing, medical imaging and radiology interpretation, clinical documentation, drug design, clinical decision and patient–provider support, synthetic data generation, and medical education and training—and maps the model families most commonly used in each theme. On the basis of this synthesis, generative AI demonstrably accelerates diagnostic workflows, augments scarce clinical datasets, personalizes communication, and supports discovery pipelines. However, the review also identifies four interlocking challenge domains that constrain clinical translation: data and security (privacy, bias, contextualization), technology and infrastructure (scalability, interpretability, vendor dependence), ethics and governance (regulation, trust, certification),and human resources (workforce skills, change management). The paper concludes with a translational path and research priorities aimed at closing these gaps. The review is intended to serve as a reference for researchers, practitioners, and policymakers working on the responsible deployment of generative AI in clinical settings. Received: 16 September 2024 | Revised: 9 May 2026 | Accepted: 1 June 2026 Conflicts of Interest The author declares that he has no conflicts of interest to this work. Data Availability Statement Data sharing is not applicable to this article as no new data were created or analyzed in this study Author Contribution Statement Wael Rahhal: Methodology, Software, Validation, Formal analysis, Investigation, Resources, Data curation, Writing – original draft, Writing – review & editing, Visualization, Supervision, Project administration.