Hierarchical reinforcement learning for automatic disease diagnosis.

Journal: Bioinformatics (Oxford, England)
Published Date:

Abstract

MOTIVATION: Disease diagnosis-oriented dialog system models the interactive consultation procedure as the Markov decision process, and reinforcement learning algorithms are used to solve the problem. Existing approaches usually employ a flat policy structure that treat all symptoms and diseases equally for action making. This strategy works well in a simple scenario when the action space is small; however, its efficiency will be challenged in the real environment. Inspired by the offline consultation process, we propose to integrate a hierarchical policy structure of two levels into the dialog system for policy learning. The high-level policy consists of a master model that is responsible for triggering a low-level model, the low-level policy consists of several symptom checkers and a disease classifier. The proposed policy structure is capable to deal with diagnosis problem including large number of diseases and symptoms.

Authors

  • Cheng Zhong
    Lawrence Berkeley National Laboratory, Berkeley CA USA.
  • Kangenbei Liao
    School of Data Science, Fudan University, 200433 Shanghai, China.
  • Wei Chen
    Department of Urology, Zigong Fourth People's Hospital, Sichuan, China.
  • Qianlong Liu
    Alibaba Group, 310052 Hangzhou, China.
  • Baolin Peng
    Microsoft Research, Redmond, WA 98052, USA.
  • Xuanjing Huang
    School of Computer Science, Fudan University, 200433 Shanghai, China.
  • Jiajie Peng
    School of Computer Science and Technology, Harbin Institute of Technology, Harbin, China. jiajiepeng@hit.edu.cn.
  • Zhongyu Wei
    School of Data Science, Fudan University, Shanghai, China.