Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Generative AI in Medicine

Divya Shanmugam, Monica Agrawal, Rajiv Movva, Irene Y. Chen, Marzyeh Ghassemi, Maia Jacobs, Emma Pierson

TL;DR

The paper addresses how generative AI can transform medicine while raising privacy, accuracy, and equity concerns. It surveys model families (e.g., large language models, diffusion models, vision-language models), delineates use cases for clinicians, patients, trial organizers, researchers, and trainees, and discusses critical challenges such as informed consent, privacy, transparency, and accountability. Key contributions include the discussion of retrieval-augmented generation to reduce hallucinations, synthetic data techniques for privacy, and the need for fine-grained real-world evaluation and governance frameworks. The work offers a roadmap for safe, effective deployment of generative AI in healthcare with implications for policy, interface design, and ongoing research.

Abstract

The increased capabilities of generative AI have dramatically expanded its possible use cases in medicine. We provide a comprehensive overview of generative AI use cases for clinicians, patients, clinical trial organizers, researchers, and trainees. We then discuss the many challenges -- including maintaining privacy and security, improving transparency and interpretability, upholding equity, and rigorously evaluating models -- which must be overcome to realize this potential, and the open research directions they give rise to.

Generative AI in Medicine

TL;DR

The paper addresses how generative AI can transform medicine while raising privacy, accuracy, and equity concerns. It surveys model families (e.g., large language models, diffusion models, vision-language models), delineates use cases for clinicians, patients, trial organizers, researchers, and trainees, and discusses critical challenges such as informed consent, privacy, transparency, and accountability. Key contributions include the discussion of retrieval-augmented generation to reduce hallucinations, synthetic data techniques for privacy, and the need for fine-grained real-world evaluation and governance frameworks. The work offers a roadmap for safe, effective deployment of generative AI in healthcare with implications for policy, interface design, and ongoing research.

Abstract

The increased capabilities of generative AI have dramatically expanded its possible use cases in medicine. We provide a comprehensive overview of generative AI use cases for clinicians, patients, clinical trial organizers, researchers, and trainees. We then discuss the many challenges -- including maintaining privacy and security, improving transparency and interpretability, upholding equity, and rigorously evaluating models -- which must be overcome to realize this potential, and the open research directions they give rise to.

Paper Structure

This paper contains 18 sections, 2 figures.

Table of Contents

  1. Introduction
  2. Background on Generative AI
  3. Use cases for generative AI in medicine
  4. Clinicians
  5. Patients
  6. Trial Organizers
  7. Researchers
  8. Trainees
  9. Challenges and directions for future work
  10. Ensuring informed consent
  11. Maintaining privacy and security
  12. Improving transparency and interpretability
  13. Mitigating hallucinations
  14. Designing usable interfaces
  15. Centering equity
  16. ...and 3 more sections

Figures (2)

  • Figure 1: We highlight promising use cases of generative AI in medicine for five key constituent groups: clinicians, patients, trial organizers, researchers, and trainees.
  • Figure 2: Bridging the gap between generative models in theory and practice will require addressing key challenges to mitigate risks and maximize benefits.