Skip to main content
AAO-HNSF 2026 Annual Meeting & OTO EXPO Main banner
Icon Legend
This session is not in your schedule.
This session is in your schedule. Click again to remove it.

Scientific Oral Presentation

Laryngology/Broncho-Esophagology

Accuracy of Artificial Intelligence in the Management of Dysphagia in Head and Neck Cancer Patients

Tuesday, October 20, 2026
8:18 AM - 8:24 AM PT
Location: 406B

    Presenting Author(s)

  • Andrea L. Jakubowski, BS

    Medical Student
    Lake Erie College of Osteopathic Medicine
    Doylestown, Pennsylvania, United States

  • SB

    Sana S. Baban, MBE

    Medical Student
    UMass Chan Medical School
    Wakefield, Massachusetts, United States

    • Co-Author(s)

  • SB

    Sana S. Baban, MBE

    Medical Student
    UMass Chan Medical School
    Wakefield, Massachusetts, United States

  • HH

    Hailey Harbert, BSN, RN

    Medical Student
    Virginia College of Osteopathic Medicine
    Princeton, West Virginia, United States

    • Senior Author(s)

  • Inna A. Husain, MD

    Medical Director of Laryngology
    Powers Health
    Chicago, IL, United States

Disclosure(s):
Andrea L. Jakubowski, BS:

faculty for this accredited education activity has no relationship(s) with ineligible companies to disclose.

Introduction: Dysphagia in head and neck cancer patients significantly impacts quality of life and often requires early identification to prevent severe complications. This study evaluates the accuracy of artificial intelligence (AI) platforms in assessing and managing dysphagia within this population.

Methods: In this cross-sectional evaluative study conducted in November 2025, 31questions based on AAO-HNS consensus statements regarding dysphagia assessment, management, and risk factors in head and neck cancer patients were presented to ChatGPT-5 and Open Evidence (OE). The primary outcomes were accuracy ratings provided by five laryngologists and five speech-language pathologists using a Likert scale. Independent variables included the AI platform type (general-purpose vs. medical-specialized) and the clinical category of query. Descriptive statistics and inter-rater reliability were calculated to compare performance across assessment, management, and risk factor identification.

Results: Mean accuracy ratings for ChatGPT-5 and OE were 4.13 vs. 4.27 for assessment, 4.25 vs. 4.31 for management, and 3.84 vs. 4.09 for risk factors, respectively. Inter-rater reliability was substantial, with a Fleiss Kappa score of 0.615. The lowest performing area was the identification of risk factors by ChatGPT-5. Statistical analysis revealed no significant difference in performance between the general-purpose generative AI (ChatGPT-5) and the specialized medical AI (OE) across the tested domains.

Conclusions: AI platforms demonstrate high accuracy in the assessment and management of dysphagia in head and neck cancer, though performance is notably lower regarding risk factor identification. While AI serves as a promising clinical adjunct, the lack of statistical difference between general and specialized platforms suggest that current public-access AI provides accuracy comparable to evidence-based medical tools, albeit with a persistent need for expert oversight in risk stratification.