Finding Optimal Cancer Treatment using Markov Decision Process to Improve Overall Health and Quality of Life

TL;DR

This paper proposes a framework using Markov Decision Processes to optimize cancer treatment strategies by considering ambient factors and overall quality of life, not just disease eradication.

Contribution

It introduces a comprehensive model that incorporates patient and physician utilities, ambient factors, and financial considerations into treatment decision-making.

Findings

Optimal treatment actions are sensitive to extraneous factors like financial status.

The framework can incorporate diverse utilities and probabilities for personalized treatment.

Simulations show improved quality of life outcomes with the proposed approach.

Abstract

Markov Decision Processes and Dynamic Treatment Regimes have grown increasingly popular in the treatment of diseases, including cancer. However, cancer treatment often impacts quality of life drastically, and people often fail to take treatments that are sustainable, affordable and can be adhered to. In this paper, we emphasize the usage of ambient factors like profession, radioactive exposure, food habits on the treatment choice, keeping in mind that the aim is not just to relieve the patient of his disease, but rather to maximize his overall physical, social and mental well being. We delineate a general framework which can directly incorporate a net benefit function from a physician as well as patient's utility, and can incorporate the varying probabilities of exposure and survival of patients of varying medical profiles. We also show by simulations that the optimal choice of actions often is sensitive to extraneous factors, like the financial status of a person (as a proxy for the affordability of treatment), and that these actions should be welcome keeping in mind the overall quality of life.