Date/Time
Date(s) - 20/11/2024
10:30 am - 11:30 am

Location
Physics and Astronomy Department

ABB 102

Speaker: Dr. Tom Purdie

Title: Associate Professor
University of Toronto
Dr. Purdie Profile Link

Title: Navigating Challenges and Opportunities in Machine Learning Automation for Radiation Oncology

Abstract:

Despite the significant potential of medical artificial intelligence (AI) to enhance standardization and efficiency in healthcare, AI technologies often face substantial barriers in transitioning to routine clinical practice. Even when AI models demonstrate validation results that match or surpass human performance, they encounter difficulties in integrating into clinical workflows due to the controlled nature of research environments compared to the dynamic complexity of clinical settings.

Key factors hindering the clinical applicability of AI technologies include the quality and representativeness of the data used to train AI models, awareness of the clinical limitations of these models, regulatory challenges specific to AI in medicine, and biases—both inherent in AI systems and in how clinicians perceive and utilize these tools. Medical physicists play a crucial role in overcoming these obstacles, ensuring that AI is applied appropriately and produces results that meet clinical standards. They serve as vital intermediaries, helping to translate AI innovation into practical clinical tools by rigorously testing models in real-world environments, guiding regulatory compliance, and managing clinical expectations.

To maximize the potential of AI in healthcare, it is essential that medical physicists not only set realistic expectations for what AI can achieve but also lead the ongoing refinement and adaptation of these technologies, balancing innovation with the practical demands of clinical practice.