Promise of Data-Driven Modeling and Decision Support for Precision Oncology and Theranostics

TL;DR

This work addresses the need for personalized theranostic strategies in cancer by proposing a data-driven framework that blends reinforcement learning with physiologically grounded models. It advances a Neural ODE–PBPK integration to capture time-varying pharmacokinetics and pairs it with an RL-based decision-support system to optimize radiopharmaceutical dosing, using an MDP formulation and neural-ODE-driven transition dynamics. Key contributions include a detailed proposal for MDP components, policy development, offline/online training strategies, and a comprehensive set of performance metrics for evaluation. The approach aims to improve tumor control while limiting radiation exposure to healthy tissue, leveraging Theranostic Digital Twin concepts and advanced simulation to bridge research and clinical practice.

Abstract

Cancer remains a leading cause of death worldwide, necessitating personalized treatment approaches to improve outcomes. Theranostics, combining molecular-level imaging with targeted therapy, offers potential for precision oncology but requires optimized, patient-specific care plans. This paper investigates state-of-the-art data-driven decision support applications with a reinforcement learning focus in precision oncology. We review current applications, training environments, state-space representation, performance evaluation criteria, and measurement of risk and reward, highlighting key challenges. We propose a framework integrating data-driven modeling with reinforcement learning-based decision support to optimize radiopharmaceutical therapy dosing, addressing identified challenges and setting directions for future research. The framework leverages Neural Ordinary Differential Equations and Physics-Informed Neural Networks to enhance Physiologically Based Pharmacokinetic models while applying reinforcement learning algorithms to iteratively refine treatment policies based on patient-specific data.