To systematically collate and interpret the applications of generative artificial intelligence (GenAI) in head and neck oncology, emphasizing its performance, risks, and ethical implications, while highlighting the systematic nature of the review.
Key Findings:
GenAI shows potential in language-centric tasks such as TNM staging and patient education but remains unreliable for autonomous clinical decision-making, as evidenced by specific studies.
The complexity of head and neck oncology presents unique challenges that GenAI may not adequately address, particularly in nuanced decision-making.
Current evidence is fragmented and heterogeneous, lacking a comprehensive synthesis specific to head and neck oncology, which limits its applicability.
Interpretation:
GenAI should be viewed as an assistive technology that augments expert clinical judgment, particularly in the complex field of head and neck oncology, by providing support in specific tasks.
Limitations:
The evidence base is heterogeneous in methodology and interpretation, which complicates the assessment of GenAI's effectiveness.
No comprehensive review has previously synthesized GenAI applications specifically for head and neck oncology, limiting understanding of its potential.
Conclusion:
A principled framework for responsible integration of GenAI in head and neck oncology is necessary to distinguish its genuine value from its limitations, suggesting specific principles for implementation.
