Machine learning solutions in radiology: does the emperor have no clothes?

Area of Science:

• Advancements in medical imaging analysis.
• Integration of artificial intelligence in healthcare.
• Development of computational pathology tools.

Background:

• Radiomics and machine learning (ML) show increasing research interest and commercial solutions.
• Growing body of research exploring ML applications in medical diagnostics.

Discussion:

• Current ML-based commercial products often have limited clinical validation.
• Methodological quality of studies supporting these tools is frequently low.
• Significant ethical and regulatory challenges hinder clinical implementation.

Key Insights:

• Despite rising interest, clinical utility of commercial ML tools is not well-established.
• Low-quality studies and lack of robust evidence characterize the current landscape.
• Ethical and regulatory concerns are major barriers to adoption.

Outlook:

• Further high-quality research is needed to validate ML tools.
• Addressing ethical and regulatory issues is crucial for clinical integration.
• Future development should focus on evidence-based, clinically relevant AI solutions.