Democratizing Artificial Intelligence Imaging Analysis With Automated Machine Learning: Tutorial

Area of Science:

• Medical Artificial Intelligence
• Machine Learning Engineering
• Clinical Informatics

Background:

• Deep learning in clinical imaging analysis powers diagnostic AI, matching or surpassing expert performance and revolutionizing healthcare.
• Automated machine learning (autoML) platforms reduce technical barriers, enabling clinicians with limited expertise to leverage AI, including foundation models like large language models.

Purpose of the Study:

• To provide a technical overview of autoML platforms and their applications in medical education, research, and clinical practice.
• To outline the stages of an autoML project, emphasizing ethical and technical best practices.
• To discuss the strengths and limitations of various autoML platforms (code-free, code-minimal, code-intensive).

Main Methods:

• Review and technical overview of autoML processes.
• Description of autoML application stages: data acquisition, partitioning, model training, validation, analysis, and deployment.
• Evaluation of different autoML platform types based on coding requirements.

Main Results:

• AutoML democratizes AI in medicine, enhancing AI literacy through hands-on education.
• AutoML facilitates rapid research testing and benchmarking, optimizing resource allocation.
• AutoML can be applied in clinical settings, pending regulatory compliance, and promotes data-centric development.

Conclusions:

• AutoML holds significant potential to advance AI adoption in medicine by simplifying complex processes.
• Effective and ethical implementation requires comprehensive education for clinicians on autoML technologies.
• AutoML supports a shift towards data-centric approaches in medical AI development.