Creation and Adoption of Large Language Models in Medicine
JAMA Network
|
Summary
Increased interest in using large language models (LLMs) in medicine could yield benefits, but their development must be shaped by relevant training data, verified benefits, and real-world evaluations. Medical LLMs can be trained on medical codes or documents, providing useful capabilities in language manipulation like summarization and translation. Evaluations should also verify purported benefits using practical, real-world situations. This professional shaping can avoid potential disruption and improve patient care.