Leveraging foundation and large language models in medical artificial intelligence

Area of Science:

• Medical Artificial Intelligence
• Machine Learning in Healthcare
• Clinical Applications of AI

Background:

• Foundational and large language models (LLMs) are increasingly prevalent in medical AI.
• These advanced AI models offer significant potential for healthcare applications.

Purpose of the Study:

• To review the applications of foundational and LLMs in medical AI.
• To introduce a novel classification framework for medical AI models.
• To address challenges and considerations for implementing AI in healthcare.

Main Methods:

• Literature review of recent advancements in medical AI.
• Development of a classification system for AI models (disease-specific, general-domain, multi-modal).
• Analysis of challenges including data acquisition, augmentation, fusion, and privacy.

Main Results:

• A new framework categorizes medical AI models into three types.
• Key challenges in data handling, model evaluation, and regulatory oversight are identified.
• The transformative potential of AI in healthcare is highlighted.

Conclusions:

• Responsible integration of AI in healthcare requires addressing data, evaluation, and regulatory challenges.
• Continuous improvement, data security, and standardized evaluations are crucial.
• Collaborative approaches are essential for the effective use of AI in clinical settings.