ICBO 2014 Proceedings An initial framework for an ontology-based mobile application for vaccine information retrieval and education Muhammad ”Tuan” Amith, MS Cui Tao, PhD School of Biomedical Informatics School of Biomedical Informatics University of Texas Health Science Center University of Texas Health Science Center Houston, Texas Houston, Texas Email: muhammad.f.amith@uth.tmc.edu Email: cui.tao@uth.tmc.edu Abstract—The Center for Disease Control (CDC) distributes Vaccine Information Statements (VIS) documentation to clinician <<:Java Web Server>> <> to inform patients of the benefits and risk of the vaccines that are Vaccine Ontology available. The effectiveness of the learning and comprehension Application has been reported to have been insufficient, particularly with Server population with lower than average literacy rates. This paper <> Fuseki introduces a proof of concept mobile application, driven by a manifests manifests vaccine ontology based on a sample set of VIS documents, as a possible solution. SPARQL Endpoint Keywords—Mobile agents, Knowledge engineering, Vaccines, Biomedical informatics I. BACKGROUND SPARQL Query JSON Response Request Object An initial literature review revealed: • Vaccine Information Statements is informally read and comprehended by patients. [1] [2] <> iOS • Interest and efforts have been made to improve the delivery of vaccine education. [1] [3] [4] Translator • Internet misinformation campaign have had significant Vaccine manifests influence on attitudes towards vaccines. [5] [6] Helmsman NLI manifests We introduce a prototype ontology-driven application system (Vaccine Helmsman) which employs a semantic knowledge-base, that harness the use of natural language inter- Fig. 1. System architecture of Vaccine Helmsman face of a mobile client for patients to query vaccine knowledge. This prototype intends to improve patient education of vaccines as an on-demand, ubiquitous aid to provide accurate and timely an overview of the system and the communication between information about vaccinations. the mobile application and the server components. The UML diagram depicts the mobile application sending a SPARQL II. A PPROACH & S YSTEM OVERVIEW [7] query request to the application server after the natural language interface (NLI) retrieves the user’s command and With the development of this prototype mobile assistant, sends it to the translator component for query construction. we plan on accomplishing the following initial aims. The server responds to the request and preforms the query on a • Specific Aim 1. Create the Vaccine Information State- published OWL2 [8] file. The server applications returns back ment Ontology (VISO). a JSON [9] data object to the mobile application. The mobile application parses the data object with the translator com- • Specific Aim 2. Vaccine Helmsman: Develop the pro- ponent, and the NLI component constructs a semi-structured totype application with client and server components. sentence to speak back to the user with an answer for the user query. The Vaccine Helmsman system employs a client-server model - manifested by the mobile device and server application Based on the architecture design described above, we hosted on the authors’ development server. Figure 1 shows have developed a working prototype that both demonstrates 81 ICBO 2014 Proceedings TABLE I. S YSTEM LIMITATIONS AND FUTURE SOLUTIONS Limitation Plan Outcome Limited vocabulary and terms recognized by sys- Utilize SKOS and existing ontologies, and integrate Recognize synonyms (& multilingual) terms for tem synonymous, consumer-level terms queries Supports for only 2 (Rotavirus and Hepatitis B) Improve VISO design with additional VIS and Better congruency with VIS and comprehensive VIS documents vaccine sources vaccine knowledge-base System constraints with offline speech recognition, Explore options for system improvement and ex- Improved usability and overall effectiveness of the latency network issues, lack of user evaluation, etc. perimentation, and investigate opportunities for application user-centered evaluation limited number of sources to generate the ontology. Only two VIS documents for Rotavirus and Hepatitis B were used for sampling, resulting in a smaller-than-ideal knowledge- base. Work is underway to improve the VISO knowledge- base by adding additional information from more VIS doc- uments and existing ontologies, and at the time of writing, the VISO design has matured and has incorporated additional VIS knowledge. Various technical constraints exist, which is expected from a prototype. This ranges from restrictions of an offline speech recognition, latency issues relating to a client- server architecture, undiscovered usability issues, and so on. Upcoming development will explore other options to improve some of these limitations, including generating a methodology to evaluate the user experience. Fig. 2. Screenshots of iOS application: Input screen (left) and response screen R EFERENCES (right). [1] Lieu TA, Glauber JH, Fuentes-Afflick E, and Lo B, “Effects of vaccine information pamphlets on parents’ attitudes,” Archives of Pediatrics & Adolescent Medicine, vol. 148, no. 9, pp. 921-925, Sep. 1994. the feasibility of the project and highlights some potential [2] M. St-Amour, M. Guay, L. Perron, P. Clment, G. Baron, G. Petit, and direction to continue to enhance this system. The prototype J. Lemaire, “Are vaccination information leaflets useful for vaccinators iOS application was developed in Objective-C (See Figure 2 and parents?,” Vaccine, vol. 24, no. 14, pp. 2491-2496, Mar. 2006. ) that accepts a defined set of voice commands and sends [3] T. C. Davis, J. A. Bocchini, D. Fredrickson, C. Arnold, E. J. Mayeaux, the command to a remote application web server . The server P. W. Murphy, R. H. Jackson, N. Hanna, and M. Paterson, “Parent application, a Jena Fuseki [10] install, is responsible for the Comprehension of Polio Vaccine Information Pamphlets,” Pediatrics, vol. queries on the published VISO OWL file and responds with 97, no. 6, pp. 804-810, Jun. 1996. a JSON object. The iOS application then constructs semi- [4] T. M. Fitzgerald and D. E. Glotzer, “Vaccine Information Pamphlets: More Information Than Parents Want?,” Pediatrics, vol. 95, no. 3, pp. structured sentences from JSON to speak back to the user. 331-334, Mar. 1995. The application also displays the text of the speech along with [5] A. Kata, “Anti-vaccine activists, Web 2.0, and the postmodern paradigm the returned JSON data for debugging purposes. An overview of tactics and tropes used online by the anti-vaccination movement,” Vaccine, vol. 30, no. 25, pp. 37783789, May 2012. Also, the authors created an initial draft of the vaccine [6] A. Kata, “A postmodern Pandoras box: Anti-vaccination misinformation information ontology based on the CDC’s Vaccine Information on the Internet,” Vaccine, vol. 28, no. 7, pp. 17091716, Feb. 2010. Statements, identified as the Vaccine Information Statement [7] ”SPARQL 1.1 Overview.” [Online]. Available: Ontology (VISO). VISO, since it is an initial draft at the http://www.w3.org/TR/sparql11-overview/. [Accessed: 15-Aug-2014]. time of development, only models the Rotavirus and Hepatitis [8] “OWL 2 Web Ontology Language Document Overview (Second Edi- B vaccine information from their respective VIS documents. tion).” [Online]. Available: http://www.w3.org/TR/owl2-overview/. [Ac- VISO was serialized in OWL2 using Protégé [11]. cessed: 01-Jul-2014]. [9] ”JSON.” [Online]. Available: http://json.org/. [Accessed: 15-Aug-2014]. [10] ”Apache Jena - Home.” [Online]. Available: https://jena.apache.org/. III. L IMITATIONS & F UTURE D IRECTION [Accessed: 15-Aug-2014]. Because it is still in the beginning stages, limitations [11] “protégé.” [Online]. Available: http://protege.stanford.edu/. [Accessed: exist with the system, which we will address as development 01-Jul-2014]. continues. Table I outlines the limitations identified, but also [12] ”SKOS Simple Knowledge Organization System - home page.” [Online]. Available: http://www.w3.org/2004/02/skos/. [Accessed: 15-Aug-2014]. describes future solutions to address them and the intended outcome. One factor is the finite vocabulary set of the system, specifically with the VISO knowledge-base, which results in semi-structured speech and limited vocabulary for query recognition. We plan on addressing this issue by utilizing Simple Knowledge Organization Systems (SKOS) [12] that will enable an ontology to add synonymous terms. SKOS may also open possibilities to utilize multi-lingual capabil- ities to the system. Another issue relating to VISO is the 82 ICBO 2014 Proceedings An initial framework for an ontology-based mobile application for vaccine information retrieval and education Muhammad “Tuan” Amith, M.S., Cui Tao, Ph.D. The School of Biomedical Informatics |The University of Texas Health Science Center at Houston Background System Architecture Future Direction Since the early 90s, healthcare providers have Because it is still in the beginning stages, limitations been mandated to provide VIS (Vaccine exist with the system, which we will address as Information Statement) from the Centers for development continues. One factor is the finite Disease Control and Prevention (CDC) to parents vocabulary set of the system, specifically with the and patients before their children or themselves VISO knowledge-base, which results in semi- receive any vaccination uptake. Despite the structured speech and limited vocabulary for query initiative, there exist issues of patients not recognition. We plan on addressing this issue by acquiring a comprehensive understanding about utilizing Simple Knowledge Organization Systems the vaccines and some evidence of doubt of the (SKOS)[12] that will enable an ontology to add safety of vaccines. In addition, a significant synonymous terms. SKOS may also open possibilities number of patients find vaccine information on the to utilize multilingual capabilities to the system. Internet, which may inevitably influence Another issue relating to VISO is the limited number of perceptions of vaccines. This paper introduces sources to generate the ontology. Only two VIS Vaccine Helmsman, an initial prototype of a documents for Rotavirus and Hepatitis B were used for mobile client that allows for natural language sampling, resulting in a smaller-than-ideal knowledge- querying of semantically-driven knowledge-base base. Work is underway to improve the VISO of vaccine information for patients. knowledge- base by adding additional information from more VIS documents and existing ontologies, and at Approach the time of writing, the VISO design has matured and A literature review revealed that CDC-mandated has incorporated additional VIS knowledge. Various VIS are seldom read or understood[1][2], and a technical constraints exist, which is expected from a need for better delivery of vaccine education[1][3] prototype. This ranges from restrictions of an offline [4]. Additionally, misinformation campaigns have speech recognition, latency issues relating to a client- influenced attitudes and decisions for uptake[5][6] server architecture, undiscovered usability issues, and [7]. so on. Upcoming development will explore other options to improve some of these limitations, including The authors introduces a prototype ontology- generating a methodology to evaluate the user driven application system (Vaccine Helmsman) experience. which employs semantic knowledge base, permitting the use of natural language interfaces References for patients to query vaccine knowledge. The 1. Lieu T, et al. Effects of Vaccine Information Pamphlets on Parents’ Attitudes. Arch Pediatr Adolesc Med. 1994; 148: 921-925. prototype aims to address the vaccine education 2. St-Amour M, et al. Are vaccination information leaflets useful for vaccinators and parents? Vaccine 24, 2006: 2491-2496. gap. 3. Fitzgerald T, Glotzer D. Vaccine Information Pamphlets: More Information Than Parents Want?. Pediatrics. March 1995 : 331-334. 4. Davis T,et.al.Parent Comprehension of Polio Vaccine Information Pamphlets. Pediatrics Vol 97,No 6,June 1996: 804-810. The authors have developed the Vaccine 5. Kata A. Anti-vaccine activists, Web 2.0, and the postmodern paradigm - An overview of tactics and tropes used online by the anti-vaccination movement. Vaccine 30, 2012: 3778-3789. Information Statement Ontology (VISO) that only 6. McRee A, Reiter P, Brewer N. Parents’ Internet use for information about HPV vaccine. Vaccine 30, 2012: 3757- 3762. 7. Kata A. A postmodern Pandora’s box: Anti-vaccination information on the Internet. Vaccine 28, 2010: 1709-1715. models Rotavirus and Hepatitis B vaccine 8. OWL 2 Web Ontology Language Document Overview (Second Edition). http://www.w3.org/TR/owl2-overview/. Published December 11, 2012. Accessed March 1, 2014. information, represented in OWL[8], using 9. Protégé. http://protege.stanford.edu/. Accessed March 10, 2014. Protégé[9]. Later development will include a 10. 11. Fuseki: serving RDF data over HTTP. https://jena.apache.org. Accessed April 24, 2014. JSON. http://json.org/. Accessed April 25, 2014. corpus of vaccine knowledge from CDC’s VIS. 12. SKOS Simple Knowledge Organization System. http://www.w3.org/2004/02/skos/. Accessed April 25, 2014. In conjunction with the creation of VISO ontology, the authors have developed a prototype iOS Limitations Plan Outcome Contact Information application in Objective-C that accepts defined Limited vocabulary and terms recognized by Utilize SKOS and existing ontologies, and Recognize synonyms (& multilingual) terms for set of voice commands and sends the command system integrate synonymous, consumer-level terms queries to a remote Java web server. The server Supports for only 2 (Rotavirus and Hepatitis B) Improve VISO design with additional VIS and Better congruency with VIS and comprehensive VIS documents vaccine sources vaccine knowledge-base For more information on this poster, please contact application, a Jena Fuseki[10] install, queries the Tuan Amith, MS at muhammad.f.amith@uth.tmc.edu, published VISO OWL file and responds with a System constraints with offline speech Explore options for system improvement and ex- Improved usability and overall effectiveness of Cui Tao, PhD at Cui.Tao@uth.tmc.edu JSON[11] object. The app then constructs semi- recognition, latency network issues, lack of user perimentation, and investigate opportunities for the application evaluation, etc. user-centered evaluation structured sentences from JSON to speak back to the user. 83