DIGITAL LIBRARY
INFORMATION RETRIEVAL BASED ON A SEARCH ENGINE WITH INVERTED INDEX FOR SOFIA (ONLINE SYSTEM FOR INTERACTIVE IMPROVEMENT OF PRIMARY CARE)
1 Telehealth Center (BRAZIL)
2 Polytech Grenoble (FRANCE)
About this paper:
Appears in: INTED2019 Proceedings
Publication year: 2019
Pages: 9279-9283
ISBN: 978-84-09-08619-1
ISSN: 2340-1079
doi: 10.21125/inted.2019.2301
Conference name: 13th International Technology, Education and Development Conference
Dates: 11-13 March, 2019
Location: Valencia, Spain
Abstract:
Introduction:
The Telehealth Brazil Networks Programme is a national action that seeks to improve the quality of care and primary care in the Unified Health System (SUS), integrating teaching and service through information technology tools, which offers conditions to promote Tele-assistance and Tele-education. Regarding to tele-assistance, the Telehealth Center of the University Hospital of the Federal University of Maranhão (NTS HU-UFMA) consists of Teleconsultation services, which provide the assistance for health actions, clinical procedures and issues related to work processes. The process of a teleconsultation starts with the solicitant asking a question in the Brazilian Telehealth platform, which will be forwarded to the professional capable of answering it to the solicitant based on scientific evidences. All this process must be done within 72 hours. SOFIA, the new teleconsultation platform, has been in use since August 2018. The development of SOFIA has enabled NTS to create new features to improve the process involved in the teleconsultations services. In order to optimize the answering-time and reduce costs on this phase, NTS has developed a new flow of teleconsultations based on the inverted index technique (tf-idf algorithm). The application of the inverted index algorithm in the platform was planned to offer a search engine inside the system to achieve different objectives: identification of duplicated questions, retrieval of existing answers to avoid resubmission and loss of resource, and so on. As a start, the retrieval of existing answers was developed, which resulted in the creation of a new flow inside the platform, speeding up the process and saving the center’s resources.

Objective:
To develop python helper scripts that basically provide a relevance search in the teleconsultations questions and answers, using the concepts of inverted index data structure. Each different script are integrated to the existing platform. Also, create a new flow for the solicitants and regulators within the full teleconsultation process.

Result:
The development of the search engine led to the creation of an additional step when solicitants request for an answer. Now, before submitting the question, SOFIA uses the search engine to find relevant answers and suggest them to the solicitant. In case one of them is suitable, the solicitant marks it as acceptable and evaluates it, finishing the process. If none of them are qualifiable, the question is submitted and sent to the regulator. During the regulation, the regulator now is able to edit it and check if the search engine finds something relevant, enabling to send the answer right the way. If at any of theses steps an existing answer is not found, the process flow goes normally. This new way of providing answers has represented 42% of center’s total teleconsultations from 20/09/2018 to 13/12/2018. In the same period, 342 were answered with the search engine, resulting in a save of R$ 13,770.00 ($3.537,57 in today’s currency). Despite the promising outcomes, the search relevance still needs better adjustments.

Conclusion:
The search engine has brought innovative application in the teleconsultation web platforms, as well as resources savings to the center. The time cost was also reduced due to the fact of not involving a more complex flow. The creation of this method has shown to be promising, even though it still needs improvements to retrieve relevant answers and result in better metrics.
Keywords:
Search engine, software development, telehealth, brazil.