Skip to main content
AAO-HNSF 2024 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.
Back

Late-Breaking Scientific Oral Presentations

Care Equity, Quality, and Safety

Session: Late-Breaking Scientific Orals: The Otolaryngology Community and Systems

Head and Neck Cancer Screening Value Enhancement Via Machine Learning to Predict Diagnoses

Monday, September 30, 2024
9:54 AM – 10:00 AM EDT
Location: MBCC, 222
Disclosure(s):
Zachary D. Urdang, MD, PhD:

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

David M. Cognetti, MD:

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

Leila J. Mady, MD, PhD, MPH:

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

Joseph M. Curry, MD:

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

Introduction: Early-stage diagnosis of cancer is paramount for favorable outcomes, and this is especially true for head and neck cancer (HNC). Despite head and neck cancer awareness and free screening events (HNCAFSEs) proven efficacy in Taiwan and India, efficacy in the United States (US) in terms of cancer detection, improved HNC outcomes, and resource utilization are questionable and hotly debated. We tested the hypothesis that machine learning (ML) can successfully predict new HNC diagnosis using ICD-10 codes with the goal to enhance HNCAFSE value, improve HNC detection rates in primary care, and ultimately improve HNC outcomes via early detection.

Methods: TriNetX is a multi-national cloud-based electronic health records (EHR) large database representing about 45% of US healthcare organizations. We leveraged TriNetX to screen a battery of common ML algorithms via PyCaret to predict incident HNC. Model features included all ICD-10 codes for a given patient translated into Phecodes and basic patient demographics.

Results: ML algorithm screening using a small cohort of 20,000 patients from the general population demonstrated that the ‘adaptive boost classifier’ demonstrated the best un-tuned performance. Training this model on a cohort of 48,000 patients and testing the model on 32k patients yielded an accuracy of 99%, precision of 20%, and positive predictive value of 20%.

Conclusions: ML can define patient cohorts with 20% ppv for HNC. This is in comparison to a HNCAFSE cancer detection rate of about 0.1%. Future studies will focus on model tuning, prospectively validation, and multi-institutional promotion to leverage this technology to enhance the value of HNCAFSEs and improve HNC patient outcomes.

    Presenting Author(s)

  • Zachary D. Urdang, MD, PhD

    Postdoc
    University of Iowa
    Flagstaff, AZ, United States

    Disclosure(s):

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

    • Co-Author(s)

  • NG

    Nicholas George-Jones, MD

    University of Iowa Hospitals and Clinics, United States

    Disclosure information not submitted.

  • MM

    Mazz Musa, BS

    PhD Candidate
    University of Iowa - Department of Computer Science, United States

    Disclosure information not submitted.

  • ZA

    Zaid Al-Qurayshi, MBChB, MPH

    Assistant professor
    University of Iowa Health Care
    Iowa city, Iowa, United States

    Disclosure information not submitted.

  • Nitin A. Pagedar, MD, MPH

    Associate Professor, Otolaryngology - Head and Neck Surgery
    University of Iowa
    Iowa City, Iowa, United States

    Disclosure information not submitted.

    • Senior Author(s)

  • Nitin A. Pagedar, MD, MPH

    Associate Professor, Otolaryngology - Head and Neck Surgery
    University of Iowa
    Iowa City, Iowa, United States

    Disclosure information not submitted.

    • Co-Author(s)

  • Elizabeth E. Cottrill, MD

    Associate Professor, Otolaryngology H&N Surgery, Program Director
    Thomas Jefferson University Hospital
    Philadelphia, PA, United States

    Disclosure information not submitted.

    • Senior Author(s)

  • Elizabeth E. Cottrill, MD

    Associate Professor, Otolaryngology H&N Surgery, Program Director
    Thomas Jefferson University Hospital
    Philadelphia, PA, United States

    Disclosure information not submitted.

    • Co-Author(s)

  • David M. Cognetti, MD

    Professor and Chair
    Thomas Jefferson Univ Hosp
    Penn Valley, PA, United States

    Disclosure(s):

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

    • Senior Author(s)

  • David M. Cognetti, MD

    Professor and Chair
    Thomas Jefferson Univ Hosp
    Penn Valley, PA, United States

    Disclosure(s):

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

    • Co-Author(s)

  • Leila J. Mady, MD, PhD, MPH

    Assistant Professor
    Johns Hopkins University
    Baltimore, MD, United States

    Disclosure(s):

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

    • Senior Author(s)

  • Leila J. Mady, MD, PhD, MPH

    Assistant Professor
    Johns Hopkins University
    Baltimore, MD, United States

    Disclosure(s):

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

    • Co-Author(s)

  • Joseph M. Curry, MD

    Professor
    Thomas Jefferson University
    Wynnewood, PA, United States

    Disclosure(s):

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

    • Senior Author(s)

  • Joseph M. Curry, MD

    Professor
    Thomas Jefferson University
    Wynnewood, PA, United States

    Disclosure(s):

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