Robust AI-enabled Simulation of Treatment Paths with Markov Decision Process for Breast Cancer Patients

Development in AI/ML-based methodologies has facilitated improvement in clinical decision making at various stages of treatment in breast cancer care. While this addresses patient needs at specific stages of treatment, the overall treatment path of a patient from a holistic standpoint has remained understudied due to challenges in accessing the relevant data. In this study, we propose to develop an AI-enabled treatment path simulation for breast cancer patients while characterizing the treatment paths as a Markov decision process (MDP). In order to avoid the limitations of healthcare records, which are often incomplete and subject to misinformation, we have leveraged clinical practice guidelines and expertise from physicians at Moffitt Cancer Center to develop the MDP. Our study of developing such an MDP, leveraging domain knowledge, contributes to improving research on treatment path simulation for breast cancer patients.