The prediction of treatment outcomes for individual patients or patient populations, including tumor control and normal tissue toxicity, has always been of tremendous importance for driving clinical practice in Radiation Oncology and at the same time has been an active area within the research community for developing new algorithms, models and analysis methods. Predicting patient outcomes and developing new evidence-based therapeutic regimes are by no means a trivial task as the complexity of the problem goes beyond simple empirical observations and requires a multidisciplinary approach that relies heavily on mathematical modelling, statistical analysis and, more recently, artificial intelligence. In this session, speakers will address the diversity of challenges encountered when attempting to predict the outcome of radiotherapy.

The first presentation focus is on the radiobiological modelling of the tumor control probability (TCP) for early-stage non-small cell lung cancer. The second presentation explores the validation of a model for normal tissue complication probability (NTCP) for acute esophageal toxicity. The third presentation demonstrates how a deep learning approach can be used to predict local regional control, the occurrence of distant metastases, and overall survival in head and neck cancer patients undergoing radiotherapy. Each of these presentations introduce novel methods and unique results with high potential for impacting methods for the analysis of clinical outcome data and ultimately improving treatment outcome predictions.

This activity is available from January 24, 2022, through 11:59 p.m. Eastern time on January 23, 2025.

This activity was originally recorded at ASTRO’s 2021 Annual Meeting, October 24-27th.