Safety-First Framework for AI-Enabled Anamnesis in Head and Neck Surgery: Evidence Synthesis from a Narrative Review

Area of Science:

• Medical Informatics
• Artificial Intelligence in Healthcare
• Head and Neck Surgery

Background:

• Artificial intelligence (AI) is increasingly explored for medical history taking (anamnesis).
• Existing research often focuses on large language models (LLMs), but a broader synthesis is needed.
• Understanding AI's role in head and neck surgery requires evaluating diverse applications beyond LLMs.

Purpose of the Study:

• To synthesize evidence on AI-enabled anamnesis, excluding LLM-only approaches.
• To identify implications and research priorities for head and neck surgery.
• To evaluate AI tools for history capture, summarization, interviewing, and triage.

Main Methods:

• A PRISMA-informed narrative review was conducted.
• Searches covered major databases (MEDLINE, Embase, Scopus, etc.) and preprint servers (medRxiv, arXiv).
• Included studies evaluated AI-supported history capture, conversational interviewing, symptom checkers, EHR integration, voice interviewing, training, and ethical considerations.

Main Results:

• Fifty studies (2014-2025) showed AI's feasibility and acceptability for pre-consultation history taking, reducing documentation burden.
• Symptom checkers and triage tools exhibited variable performance and safety concerns, requiring conservative escalation strategies.
• LLM-based dialogue showed promise in controlled settings, but real-world validation and integration are limited.

Conclusions:

• AI-enabled anamnesis tools are heterogeneous with inconsistent evidence.
• Near-term head and neck surgery applications include structured intake, clinician summarization, and training.
• Autonomous triage requires rigorous, specialty-specific validation and governance before clinical adoption.