Learning a Diagnostic Strategy on Medical Data With Deep Reinforcement Learning

Mengxiao Zhu,Haogang Zhu

doi:10.1109/access.2021.3087493

Abstract

In recent years, Artificial Intelligence based disease diagnosis has drawn considerable attention both in academia and industry. In medical scenarios, a well-trained classifier can effectively detect a disease with sufficient features associating with medical tests. However, such features are not always readily available due to the high cost of time and money associating with medical tests. To address this, this study identifies the diagnostic strategy learning problem and proposes a novel framework consisting of three components to learn a diagnostic strategy with limited features. First, as we often encounter incomplete medical records of the patients, a sequence encoder is designed to encode any set of information in various sizes into fixed-length vectors. Second, taking the output of the encoder as the input, a feature selector based on reinforcement learning techniques is proposed to learn the best feature sequence for diagnosis. Finally, with the best feature sequence, an oracle classifier is used to give the final diagnosis. To evaluate the performance of the proposed method, experiments are conducted on nine real medical datasets. The results suggest that the proposed method is effective for providing personalized diagnostic strategies and makes better diagnoses with fewer features compared with existing methods.

Highlights

In recent years, Artificial Intelligence based disease diagnosis has received increasing attention both in academia and industry
This paper studies the problem of learning a diagnostic strategy on incomplete medical data, which uniquely captures the core characteristics of the medical scenarios
This study formulates the problem as a constrained optimization problem and proposes a framework that follows the idea of reinforcement learning, where each element in the Markov Decision Process (MDP) is defined and three components, consisting of the feature selector, the oracle classifier, and the sequence encoder, are designed

Summary

Introduction

Artificial Intelligence based disease diagnosis has received increasing attention both in academia and industry. Considering the process of medical inquiry, given few symptoms of a new patient, the doctor requires the patient to conduct some related medical tests, whose corresponding results would increase or decrease the doctor’s confidence of the diagnosis, and the doctor would require the patient to conduct other medical tests until he/she has enough confidence to make the final diagnosis. During the process, it costs a certain amount of money and time to measure the feature values. It is of significance to form a final diagnosis rapidly with limited medical tests

Objectives

Methods

Findings

Conclusion