Back to sessions
Sep 28
Education

EDU 34 - Generative AI in Radiation Oncology: Moving from Hype to Clinical Implementation

05:00pm - 06:00pm ET

MODERATOR(S)

Danielle Bitterman, MD - Brigham and Women's Hospital/Dana-Farber

session DESCRIPTION

Radiation oncology has spent the last decade refining task-based automation, such as auto-segmentation and dose prediction. However, the rapid emergence of Generative AI (GenAI) and Large Language Models (LLMs) offers a fundamentally new capability: the ability to synthesize information and orchestrate clinical workflows. This session moves beyond the media hype to examine concrete clinical applications of GenAI. We will explore how LLMs can synthesize physician intent from multimodal data during the consult, assist in "explainable" treatment planning optimization, and streamline post-treatment surveillance through automated signal detection. Crucially, the session balances innovation with responsibility. We will debate the unique quality assurance challenges posed by non-deterministic models, where the output may vary and propose rigorous human-in-the-loop frameworks. Attendees will leave with a practical understanding of where GenAI is justified now, where it requires prospective validation, and how to navigate the emerging landscape for these advanced tools.

learning objectives

  1. Differentiate between standard deterministic automation (e.g., rigid scripts) and probabilistic Generative AI to identify appropriate, high-value use cases within the clinical workflow.
  2. Evaluate the capability of LLMs to reduce administrative burden by synthesizing consult notes, helping with RT intent generation, operationalizing planning objectives and triaging patient-reported outcomes.
  3. Determine necessary quality assurance protocols and human-in-the-loop guardrails to ensure patient safety when implementing non-deterministic AI tools in the clinic.

Credits

AMA PRA Category 1 Credits: 1.00

Rate This Session

Presentations