Strategic cancer therapy planning: optimizing treatment and quality of life with Markov decision processes

In managing the progression of diseases, particularly cancer, Markov decision processes (MDP) and dynamic therapy regimes are gaining prominence. Despite this, cancer treatments often negatively impact patients' quality of life, leading many to abandon effective, accessible, and affordable therapies.

This paper introduces a novel MDP-based mathematical framework for optimizing multi-therapy treatment schedules in malignancy therapy. Through practical illustrations, we demonstrate the utility and applicability of the proposed framework. Our approach integrates both patient utility and the physician's net benefit function, accounting for treatment options and survival probabilities across diverse clinical profiles. The system state in our MDP model is defined by tumor progression and normal tissue side effects, while the response field encompasses treatment outcomes categorized into recurrence, tumor regression, and healthy tissue safety. At each decision stage, the physician assesses the patient's condition and selects the optimal treatment strategy to maximize the final reward, determined by the patient's health at the end state.

This framework offers a holistic approach to improving overall treatment outcomes while recognizing the importance of preserving patients' quality of life.