Federated Learning in Endodontics: A Framework for Privacy-Preserving Multicentre Artificial Intelligence.

Area of Science:

• Artificial Intelligence in Dentistry
• Medical Informatics
• Machine Learning in Endodontics

Background:

• High-quality AI models in endodontics require diverse, well-annotated datasets.
• Centralized AI training faces legal and ethical challenges due to data protection regulations.
• Federated learning (FL) offers a privacy-preserving framework for collaborative AI development.

Purpose of the Study:

• Introduce federated learning (FL) as a solution for collaborative AI in endodontics.
• Explore the principles, applications, and challenges of FL in endodontic AI.
• Outline implementation pathways and research priorities for FL in endodontics.

Main Methods:

• Narrative review of literature up to April 2026.
• Comprehensive search strategy across multiple databases.
• Comparative analysis of FL fundamentals and applications for diagnostic and decision-support tasks.

Main Results:

• FL allows institutions to retain local patient data while contributing model updates.
• Exploration of FL principles, privacy mechanisms, architectures, and challenges.
• Proposed roadmap for FL implementation: pilot studies, data standardization, clinical validation, and regulatory engagement.
• FL enhances AI development in endodontics, safeguarding patient privacy for improved diagnostics and personalized care.

Conclusions:

• Federated learning can advance AI integration in endodontics, prioritizing patient privacy.
• FL enables multicenter AI model development without data sharing, improving accuracy and generalizability.
• This approach holds potential for enhanced endodontic diagnosis, treatment planning, and outcome prediction.