A popular paradigm for data-based research in Medical Physics is to connect observable features with measured outcomes through a purely statistical model (a so called “black box”). While this method has been successful for many applications, there is the limitation that when the model predicts something unexpected, one cannot ask how the model made its prediction or be confident in predictions made outside of its training domain. In this session, we will introduce the concept of biomathematical modeling, where the driving biological mechanism are explicitly modeled rather than using a black box. These models are simple enough to design and calibrate with clinical data or experiments but are powerful enough to reveal complex behavior. With biomathematical modeling, one can not only make predictions but also interrogate how the model made its prediction. Furthermore, a well calibrated model can also be used to ask interesting clinical questions on its own using “in-silico” trials. While randomized clinical trials are indeed the gold standard for clinical knowledge, “in-silico” trials based on mathematical modeling can aid in their design, given the vastness of possible drug-radiation delivery combinations and the currently limited amount of data available for true machine learning.
Looking towards the future, as radiotherapy approaches its limits to physically conform delivered dose and biologic based therapies continue to thrive, biomathematical modeling could be a new exciting avenue for medical physicists to use their unique knowledge of physical modeling and clinical oncology. Additionally, it is clear that each patient responds uniquely to radiation, both in terms of response to the tumor and normal tissue but also regarding immunologic priming. Predicting these treatment responses will require an understanding of the biological drivers of radiation response and how these therapies interact with spatiotemporal radiation delivery, a problem at the core of biomathematical modeling in radiotherapy. In this session three speakers will deliver presentations on unique applications of biomathematical modeling to clinical radiotherapy to personalize and synergize radiation therapy as part of a complex biological system in individual patients.
Speaker 1: Shifting the Paradigm in Radiation Planning From Constraint Satisfaction to True Optimization: The Role of Mathematical Biology
Speaker 2: Modeling Radiotherapy in Combination With Targeted Agents
Speaker 3: Nanoparticle Drone Models for Delivery of Immunotherapy To Boost the Abscopal Effect
Specific Learning Objectives:
1. Introduce the concept of mechanistic biological modeling (versus statistical “black box” models) and utility of in-silico studies for hypothesis development and testing.
2. Learn about connecting biomathematical models to both clinical and experimental data, specifically the ideas of robustness and sensitivity to model parameters.
3. Discuss novel and contemporary applications of biomathematical modeling in the clinic bridging radiotherapy and new biological concepts.
Funding Support, Disclosures, and Conflict of Interest: JC is co-founder and hold shares of Cvergenx, company that holds license to RSI/GARD; CG reports the following relevant funding from NIH: R21CA248118, R21CA241918, P01CA261669; WN none
Not Applicable / None Entered.
Not Applicable / None Entered.