Dissertations & Theses (Open Access)

Graduation Date

Summer 8-14-2019

Degree Name

Doctor of Philosophy (PhD)

School Name

The University of Texas School of Biomedical Informatics at Houston

Advisory Committee

Cui Tao, PhD

Abstract

Speech and conversational technologies are increasingly being used by consumers, with the inevitability that one day they will be integrated in health care. Where this technology could be of service is in patient-provider communication, specifically for communicating the risks and benefits of vaccines. Human papillomavirus (HPV) vaccine, in particular, is a vaccine that inoculates individuals from certain HPV viruses responsible for adulthood cancers - cervical, head and neck cancers, etc. My research focuses on the architecture and development of speech-enabled conversational agent that relies on series of consumer-centric health ontologies and the technology that utilizes these ontologies. Ontologies are computable artifacts that encode and structure domain knowledge that can be utilized by machines to provide high level capabilities, such as reasoning and sharing information. I will focus the agent’s impact on the HPV vaccine domain to observe if users would respond favorably towards conversational agents and the possible impact of the agent on their beliefs of the HPV vaccine. The approach of this study involves a multi-tier structure. The first tier is the domain knowledge base, the second is the application interaction design tier, and the third is the feasibility assessment of the participants. The research in this study proposes the following questions:

  1. Can ontologies support the system architecture for a spoken conversational agent for HPV vaccine counseling?
  2. How would prospective users’ perception towards an agent and towards the HPV vaccine be impacted after using conversational agent for HPV vaccine education?

The outcome of this study is a comprehensive assessment of a system architecture of a conversational agent for patient-centric HPV vaccine counseling. Each layer of the agent architecture is regulated through domain and application ontologies, and supported by the various ontology-driven software components that I developed to compose the agent architecture. Also discussed in this work, I present preliminary evidence of high usability of the agent and improvement of the users’ health beliefs toward the HPV vaccine. All in all, I introduce a comprehensive and feasible model for the design and development of an open-sourced, ontology-driven conversational agent for any health consumer domain, and corroborate the viability of a conversational agent as a health intervention tool.

Keywords

ontology evaluation, natural language processing, Vaccine Misinformation Ontology (VAXMO), HPV, Patient Health Information Dialogue Ontology (PHIDO)

Download

Share

COinS