The new paradigm in machine learning - foundation models, large language models and beyond: a primer for physicians

Area of Science:

• Artificial Intelligence
• Machine Learning
• Medical Informatics

Background:

• Foundation models represent a paradigm shift from task-specific AI.
• Large Language Models (LLMs) like ChatGPT are prominent text-based foundation models.
• LLMs have the potential to revolutionize medical tasks and clinical decision-making.

Purpose of the Study:

• To review different types of LLMs.
• To explore emerging applications of LLMs in medicine.
• To discuss limitations, biases, and future clinical translation of LLMs.

Main Methods:

• This is a narrative review.
• The review synthesizes current knowledge on LLMs in medicine.
• It covers LLM types, applications, risks, and future directions.

Main Results:

• LLMs can automate tasks like summarizing medical records and answering patient queries.
• Potential risks include harm due to inadequate development, evaluation, and scrutiny.
• Bias and limitations within LLMs require careful consideration.

Conclusions:

• LLMs show significant promise for transforming healthcare delivery.
• Rigorous oversight is essential for the safe and effective implementation of LLMs in clinical practice.
• Further research and development are needed for responsible translation into medicine.