JMIR Publications

ABSTRACT

Background:

The development of human speech recognition technologies has made increasing progress in recent years. Their role is promising in relation to health information dissemination to facilitate the role of health workers in social distancing by distributing information to the population. Some experiments were developed during the pandemic because traditional channels (e.g. telephone help desk, websites with FAQs, social media) were compromised by the large number of user accesses. For this type of situation, it is necessary to relate the technological development paths to the contexts within which the experiments were carried out.

Objective:

The paper highlights the text-based experience of a chatbot aimed at meeting the information needs of citizens of a northern region of Italy during the second phase of COVID-19 pandemic emergency. Since 15 April 2021, the chatbot has been accessible from every page of the website of the local Health Authority. Users can type free-form questions and obtain relevant up-to-date answers distilled from different sources of information maintained by the national and provincial authorities.

Methods:

The paper makes its contribution through a mix method research design. The work proposed a retrospective analysis of the engineering processes and the technicalities of setting up the chatbot and building up the wealth of information that became the basis for its responses. The accounts come directly from the working team and the interactions with the various stakeholders involved.

Results:

The work highlights how a technology that has arisen within bureaucratic organisational contexts cannot easily draw from the various sources of information, such as news agencies, as the chatbot should provide exclusively trusted content approved by the Health Authority so some questions cannot be answered for lack of authoritative information.

Conclusions:

The study is relevant for three reasons: firstly, it highlights how AI cannot disregard the contexts of use in which it is developed; secondly, it shows that, through rigorous methods, it is possible to plan a trajectory bringing organisations from providing information designed for static channels towards interactive ones. Finally, it shows that a carefully developed strategy to provide thrusted information can generate opportunities for digital transformation in healthcare organisations.