Spoken Dialogue Systems for Medication Management

Joan Zheng, Raymond Finzel, Serguei Pakhomov, Maria Gini

Research output: Chapter in Book/Report/Conference proceedingChapter

1 Scopus citations

Abstract

The interest towards spoken dialogue systems has been rapidly growing in the last few years, including the field of health care. There is a growing need for automated systems that can do more than order airline and movie tickets, find restaurants and hotels, or find information on the internet. Eliciting information from patients about their current health and medications using natural language at the point of care is a task currently performed by skilled nurses during the intake interview in both inpatient and outpatient settings. This routine task lends itself well to automation and a well-crafted dialogue system with state management can enable standardized yet individually tailored interactions with the patient using natural language. The need for extensive domain knowledge (e.g. medications, dosages, disorders, symptoms, etc.) in order to achieve broad coverage makes this task particularly challenging. In this project, we explore the use of the PyDial framework and a medication-oriented knowledge base containing information from RxNorm to create a dialogue system capable of eliciting medication history information from patients.

Original languageEnglish (US)
Title of host publicationStudies in Computational Intelligence
PublisherSpringer Verlag
Pages119-127
Number of pages9
DOIs
StatePublished - 2020

Publication series

NameStudies in Computational Intelligence
Volume843
ISSN (Print)1860-949X
ISSN (Electronic)1860-9503

Bibliographical note

Funding Information:
Acknowledgements Work supported in part by CRA-W Distributed Research Experiences for Undergraduates program.

Publisher Copyright:
© Springer Nature Switzerland AG 2020.

Copyright:
Copyright 2019 Elsevier B.V., All rights reserved.

Fingerprint Dive into the research topics of 'Spoken Dialogue Systems for Medication Management'. Together they form a unique fingerprint.

Cite this