Guidelines for Developing and Reporting Machine Learning Predictive Models in Biomedical Research: A...

Area of Science:

• Biomedical research
• Data science
• Clinical informatics

Background:

• Machine learning (ML) models are increasingly used in biomedical research for big data analysis.
• The complexity of ML methods can lead to misuse and insufficient reporting in publications.
• Lack of standardized reporting hinders the assessment of ML model validity and interpretation of results.

Framework:

• Develop guidelines for the application and reporting of ML predictive models in clinical settings.
• Ensure correct application and sufficient reporting to distinguish true discoveries from coincidental findings.
• Establish a consensus among experts through a Delphi method involving ML specialists, clinicians, and statisticians.

Implementation:

• A multidisciplinary panel utilized an iterative Delphi method for guideline development.
• The process resulted in a comprehensive set of guidelines.
• Guidelines include essential reporting items for research articles and practical steps for developing ML predictive models.

Implications:

• Generated guidelines facilitate the correct application and consistent reporting of ML models in biomedical research.
• These guidelines aim to accelerate the adoption of big data analysis and ML methods in the biomedical field.
• Promotes reliable assessment and interpretation of ML model outputs, fostering trust in big data-driven discoveries.