Assessing the Reporting Quality of Machine Learning Algorithms in Head and Neck Oncology

Area of Science:

• Oncology
• Medical Informatics
• Artificial Intelligence

Background:

• Machine learning (ML) is increasingly used in head and neck oncology.
• Standardized reporting of ML algorithms is crucial for clinical translation and research integrity.

Purpose of the Study:

• To assess the reporting quality of ML algorithms in head and neck oncology literature.
• To evaluate adherence to the Transparent Reporting of a Multivariable Prediction Model for Individual Prognosis Or Diagnosis - Artificial Intelligence (TRIPOD-AI) criteria.

Main Methods:

• A systematic literature search was performed across major databases (PubMed, Scopus, Embase, Cochrane).
• Two independent reviewers assessed studies against the 65-point TRIPOD-AI checklist.
• Evidence levels were evaluated using the Oxford Centre for Evidence-Based Medicine (OCEBM) framework.

Main Results:

• Current reporting of ML algorithms in head and neck oncology is suboptimal.
• Key areas for improvement include dataset descriptions, model performance reporting, and sharing of models, data, and code.
• Adherence to TRIPOD-AI criteria is necessary for standardized reporting.

Conclusions:

• Inadequate reporting of ML algorithms impedes clinical application, reproducibility, and trust.
• Open access to ML models, code, and data is vital for community critique and iterative improvement.
• Enhanced reporting practices will improve model accuracy, mitigate bias, and foster clinician trust.