Background: A conversational agent is a computer programme which can interact with people using natural language. Conversational agents can feature speech, like consumer-oriented virtual assistants; or text-based conversations, such as frame-based textual dialogue systems, broadly known as chatbots. People can interact with chatbots using for example websites or instant messengers, such as WhatsApp.
Goals: This research project will design and research a chatbot, based on an instant messenger, to support young Nigerian adults living with HIV. The chatbot will, like a human companion, remind clients of medication and clinical appointments. It will also provide information and respond to questions that these young adults may have about HIV. The third function is the provision of socio-emotional support and advice.
Consortium: The project will be carried out by a consortium of Swiss and Nigerian researchers, including IT experts from the University of Applied Sciences and Arts Northwestern Switzerland and global health specialists from the University of Ibadan and from the Swiss TPH. The consortium brings together expertise in public health and HIV, Artificial Intelligence, human–computer interaction and technology-enhanced coaching and learning, and gender.
Development and evaluation: The chatbot will be developed in a way in which future users will be involved from the very beginning, to ensure that the bot will incorporate their informational and support needs. There will be several rounds of testing, evaluations, and improvements. In the final evaluation, a group of young Nigerian HIV clients will use the chatbot for several months. The outcomes in terms of wellbeing, mental and physical health and HIV knowledge will be compared with a group that will not work with the chatbot. The different ways in which the clients will use and perceive the chatbot will be further explored using interview studies. Particular attention will be also paid to the question of whether, and if so, how, gender can influence the use, perception and effects of chatbots.
Impact: Conversational agents are expected to be a crucial way of communicating with computers in the future. They will likely have a particular relevant role in global and public health, where personalised information and support could be offered to clients and patients. However, whereas chatbots are increasingly researched in high income countries, very little is known about their potential in low- and middle-income settings. In addition to the practical goals, this is one of the knowledge gaps that this project seeks to address.
Background, rationale and goals: Artificial intelligence (AI)-based agents such as chatbots are becoming cornerstones of human-computer interaction. Yet the understanding of these technologies is, according to systematic reviews, still limited. This holds particularly for newer hybrid-adaptive chatbots, which permit more flexible, human-like conversations. The proposed project sets out to design, test, improve, and comprehensively evaluate an intelligent chatbot which acts as a personal health coach. The bot will function similarly to a human coach. It will support young Nigerian adults living with HIV by (1) reminding them of medication and clinical appointments, (2) providing information and responding to questions about HIV, and (3) offering socio-emotional support and coaching based on cognitive behavioural principles. For the planned trial, the chatbot will be integrated in a secure messaging app, but the technology will be extendable to other mobile media and web-based applications. Approach and methodsThe project will be carried out by a transdisciplinary consortium of Swiss and Nigerian researchers with expertise in public health and HIV, gender, Artificial Intelligence, human-computer interaction and technology-enhanced coaching and learning, from the Intelligent Systems Group at FHNW, the health promotion Department from the College of Medicine, University of Ibadan, and the Swiss TPH as a partner. The bot development will be based on participatory, design science research methods with iterative rounds of testing and improvement. The evaluation will involve a qualitative study and a two-arm (1:1), open label, randomised controlled superiority trial (RCT). The trial will examine the bot’s use, usability, and effects in terms of mental health and well-being, cognitive variables (knowledge, treatment self-efficacy), physiological health (viral load), and gender aspects, by comparing the use of an ‘intelligent’, hybrid-adaptive chatbot with a control condition that receives the standard of care (SoC). Repeated qualitative interviews will be used to explore the multifaceted experiences of HIV clients in using the chatbot in their daily lives, as well as gender differences. Expected findings and relevanceThe project, which will be supported by an advisory board, is expected to make a significant contribution by: designing and researching a hybrid-adaptive bot based on conversational AI in the field of HIV/AIDS in a Sub-Saharan African context, using a rich and rigorous RCT-based design; exploring use and effects over an extended period of time, and generating an in-depth understanding of the phenomenon including the exploration of gender differences. All these aspects have not or only rarely been considered in previous work. In this way, the proposed research will break new ground not only regarding digital HIV-treatment support interventions; it will also inform other domains of (digital) health, conversational AI, and technology-enhanced coaching, and we see it is a jumping off point to further scrutinise the use of conversational AI in public health settings in Sub-Saharan Africa. The project’s immediate practice relevance is also noteworthy, with HIV/AIDS presenting a very serious public health problem. Moreover, the project will consider gender aspects in the design and research of the chatbot and will methodically facilitate capacity development of the young female researchers involved.
Last updated:27.01.2023
