Tweets by AAO-HNSF 2024 Annual Meeting & OTO EXPO

Icon Legend

This session is not in your schedule.

This session is in your schedule. Click again to remove it.

Back

Favorite

Scientific Oral Presentations

Care Equity, Quality, and Safety

Session: Scientific Oral Presentations 1 - Care Equity, Quality, and Safety

Machine-Learning Methods in Classification of Prolonged Radiation Therapy in Oropharyngeal Cancer

Sunday, September 29, 2024

1:18 PM – 1:24 PM EDT

Location: MBCC, 224

Disclosure(s):

Alexis H. Kim:

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

Eun Jeong Oh, PhD:

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

Introduction: The literature has shown that certain demographic groups may be at high-risk for having treatment delays in oropharyngeal squamous cell carcinoma (OPSCC), thus decreasing care equity. These high-risk groups were previously identified based on traditional regression algorithms. Machine learning (ML) methods may have the potential to more accurately identify patients who are high-risk of having treatment delays.

In this context, the hypothesis of this study was to investigate the accuracy of ML algorithms, compared with traditional algorithms in stratifying risk of treatment delay for patients with OPSCC and to correlate with survival differences in risk-stratified groups.

Methods: The National Cancer Database was queried between 2004 to 2016 for patients with OPSCC treated with radiation therapy (RT) or chemoradiation as primary treatment. To predict risk of treatment delay, 8 different ML algorithms were compared against traditional logistic regression using various performance metrics. Data was split into a distribution of 70% for training and 30% for testing. Overall survival (OS) of risk stratified groups was assessed through Kaplan-Meier curves by human papillomavirus (HPV) status, tumor size, race, and other clinicopathologic factors.

Results: A total of 3,152 patients were included (1,928 prolonged RT, 1,224 not prolonged RT). As a whole, based on performance metrics, random forest (RF) was found to most accurately predict prolonged RT compared to both other ML methods and traditional logistic regression. RF-stratified risk groups demonstrated a difference in OS among HPV+, T3 tumors (P = 0.044) based on Kaplan-Meier curves. A difference in overall survival in stratified groups among Black patients in HPV- cases was also detected (P = 0.026).

Conclusions: Our assessment of various ML techniques showed that RF was superior to traditional logistic regression at classifying OPSCC patients at risk of treatment delay. Application of such algorithms may have potential to identify high-risk patients and enable early interventions to improve survival and care equity in OPSCC.

Presenting Author(s)

Alexis H. Kim, BA

Medical Student
The Donald and Barbara Zucker School of Medicine at Hofstra/Northwell
Hempstead, NY, United States

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

Co-Author(s)

Matthew I. Saleem, BS

Medical Student
Donald and Barbara Zucker School of Medicine at Hofstra/Northwell
Hempstead, New York, United States

Disclosure information not submitted.
EO

Eun Jeong Oh, PhD

Assistant Professor, Quantitative Intelligence
Feinstein Institutes for Medical Research
New York, NY, United States

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

Seungjun Ahn, PhD

Assistant Professor
Department of Population Health Science and Policy, Icahn School of Medicine at Mount Sinai, New York, United States

Disclosure information not submitted.

Senior Author(s)

Tristan M. Tham, MD

Assistant Professor - Otolaryngology Head & Neck Surgery
Stanford University
Stanford, CA, United States

Disclosure information not submitted.
SA

Seungjun Ahn, PhD

Assistant Professor
Department of Population Health Science and Policy, Icahn School of Medicine at Mount Sinai, New York, United States

Disclosure information not submitted.