Frontiers in Oncology (May 2024)

Assessing the role of advanced artificial intelligence as a tool in multidisciplinary tumor board decision-making for primary head and neck cancer cases

  • Benedikt Schmidl,
  • Tobias Hütten,
  • Steffi Pigorsch,
  • Fabian Stögbauer,
  • Cosima C. Hoch,
  • Timon Hussain,
  • Barbara Wollenberg,
  • Markus Wirth

DOI
https://doi.org/10.3389/fonc.2024.1353031
Journal volume & issue
Vol. 14

Abstract

Read online

BackgroundHead and neck squamous cell carcinoma (HNSCC) is a complex malignancy that requires a multidisciplinary approach in clinical practice, especially in tumor board discussions. In recent years, artificial intelligence has emerged as a tool to assist healthcare professionals in making informed decisions. This study investigates the application of ChatGPT 3.5 and ChatGPT 4.0, natural language processing models, in tumor board decision-making.MethodsWe conducted a pilot study in October 2023 on 20 consecutive head and neck cancer patients discussed in our multidisciplinary tumor board (MDT). Patients with a primary diagnosis of head and neck cancer were included. The MDT and ChatGPT 3.5 and ChatGPT 4.0 recommendations for each patient were compared by two independent reviewers and the number of therapy options, the clinical recommendation, the explanation and the summarization were graded.ResultsIn this study, ChatGPT 3.5 provided mostly general answers for surgery, chemotherapy, and radiation therapy. For clinical recommendation, explanation and summarization ChatGPT 3.5 and 4.0 scored well, but demonstrated to be mostly an assisting tool, suggesting significantly more therapy options than our MDT, while some of the recommended treatment modalities like primary immunotherapy are not part of the current treatment guidelines.ConclusionsThis research demonstrates that advanced AI models at the moment can merely assist in the MDT setting, since the current versions list common therapy options, but sometimes recommend incorrect treatment options and in the case of ChatGPT 3.5 lack information on the source material.

Keywords