Reinforcement Learning for Personalized Drug Discovery and Design for Complex Diseases: A Systems Pharmacology Perspective

  • 2022-01-21 21:29:46
  • Ryan K. Tan, Yang Liu, Lei Xie
  • 0

Abstract

Many multi-genic systematic diseases such as Alzheimer's disease and majorityof cancers do not have effective treatments yet. Systems pharmacology is apotentially effective approach to designing personalized therapies foruntreatable complexed diseases. In this article, we review the potential ofreinforcement learning in systems pharmacology-oriented drug discovery anddesign. In spite of successful application of advanced reinforcement learningtechniques to target-based drug discovery, new reinforcement learningtechniques are needed to boost generalizability and transferability ofreinforcement learning in partially observed and changing environments,optimize multi-objective reward functions for system-level molecular phenotypereadouts and generalize predictive models for out-of-distribution data. Asynergistic integration of reinforcement learning with other machine learningtechniques and related fields such as biophysics and quantum computing isneeded to achieve the ultimate goal of systems pharmacology-oriented de novodrug design for personalized medicine.

 

Quick Read (beta)

loading the full paper ...