A new, open-access JAMA Health Policy essay, "How to Safely Integrate Large Language Models Into Health Care," two authors with both industry and FDA experience discuss pathways for integration of Large Language Models (LLMs) in healthcare. Dr. Scott Gottlieb served as FDA commissioner and Lauren Silvis is a former senior FDA regulator now at TEMPUS.
The essay highlights the potential of LLM's to enhance patient interactions and healthcare delivery. Three takeaways:
LLM Potential: The essay underscores LLMs' transformative potential in healthcare, enabling human-like text generation, aiding in diagnosis, and providing ongoing patient support.
Staged Integration: The authors advocate a cautious and phased approach, starting with well-understood conditions, to help ensure safe LLM integration and boost confidence.
Data and Bias: Effective LLM use in healthcare demands improved data sharing, bias detection, and inter-institutional collaboration for accurate and unbiased results.