Predictive Digital Twins for Optimizing Patient-Specific Therapies
Graham Pash, University of Texas at Austin
Heterogeneity in high-grade glioma response to standard-of-care radiotherapy contributes to suboptimal patient outcomes. Patient-specific adaptive regimens have promise for improving outcomes by accounting for the diverse nature of disease physiology and therapeutic response. However, only a limited number of adaptive strategies based on large population-based clinical studies are considered in practice. To move treatment to a more anticipatory and personal framework, we must leverage physics-based mechanistic models and account for the underlying uncertainty in imaging, segmentations, and other data that inform clinical decision making. Predictive digital twins provide a rigorous framework for dynamically integrating data from a physical system with mathematical and computational models to predict behavior and enable decision-making. Formally, predictive digital twins employ a probabilistic graphical model of the interactions between six key elements: the physical state, the digital state, observational data, control inputs, quantities of interest, and rewards. In the context of oncology, we wish to leverage a digital twin of the patient's tumor to forecast response under both standard and clinically-feasible adaptive radiotherapy plans to guide real-time therapeutic decision making under uncertainty.
Abstract Author(s): Graham Pash, Karen Willcox, Anirban Chaudhuri, David Hormuth, Ernesto Lima, Guillermo Lorenzo, Chengyue Wu, Thomas Yankeelov