Scholarlink

This chapter shows the use of the Q-learningQ-learning framework presented in Sect.  2.8 to address the problem of rationalizing the drug injection feedback strategy in the combined (Chemotherapy/Immunotherapy) ofCancer cancer. The objective here is to induce a decrease (contraction) in the tumor size while maintaining the level of lymphocytes population above some health-defined threshold as the level of this population of cells reflects the patient’s state of resistance and health. The results of this chapter emphasize the relevance of using a non zero weighting on the variance of the cost functionCost function (augmented by the exact penalty on the constraints) in order to handle the risk management that is mandatory in this kind of applications. This terms is rarely used in the applications which mainly focus on economic criteria in which only the expectation is relevant. This is obviously not relevant in the case ofCancer cancer treatment as a loss for a patient cannot be compensated by a success on another patient. While the general framework follows the guidelines recalled in Chap.  2 , it is shown that the specific mathematical structure of the equations governing theCancer cancer/drug dynamics is exploited in order to come out with a more tractable heuristic in terms of the evaluation of the statistical moments involved in the definition of theCost function cost function.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Explicit Q-Learning-Based Approximation of Stochastic Optimal Control Feedback for Combined Therapy of Cancer

  • 章节

  • 发表日期：

    2026年

查看全文
  • Mazen Alamir

摘要

This chapter shows the use of the Q-learningQ-learning framework presented in Sect.  2.8 to address the problem of rationalizing the drug injection feedback strategy in the combined (Chemotherapy/Immunotherapy) ofCancer cancer. The objective here is to induce a decrease (contraction) in the tumor size while maintaining the level of lymphocytes population above some health-defined threshold as the level of this population of cells reflects the patient’s state of resistance and health. The results of this chapter emphasize the relevance of using a non zero weighting on the variance of the cost functionCost function (augmented by the exact penalty on the constraints) in order to handle the risk management that is mandatory in this kind of applications. This terms is rarely used in the applications which mainly focus on economic criteria in which only the expectation is relevant. This is obviously not relevant in the case ofCancer cancer treatment as a loss for a patient cannot be compensated by a success on another patient. While the general framework follows the guidelines recalled in Chap.  2 , it is shown that the specific mathematical structure of the equations governing theCancer cancer/drug dynamics is exploited in order to come out with a more tractable heuristic in terms of the evaluation of the statistical moments involved in the definition of theCost function cost function.