Development of Chatbot Technology for Acquiring and Analysing Views of Patients for Use in Healthcare Interventions

With recent developments in AI, there is an increasing interest in chatbot technology. A chatbot is not intended to fully understand natural language, but rather works with patterns frequently used in language to participate in a dialogue that is engaging and useful. In this project, we want to develop chatbot technology as part of a toolkit for acquiring and analysing views of people on healthcare issues. Consider for example, in behaviour change, different people have different reasons for not eating a healthy diet or for not undertaking exercise, and the aim of the chatbot is to discuss these issues with different people in order to acquire data on the range of reasons. The chatbot will ask questions and ask for clarifications of the user according to a dynamic script that is based on features (e.g. keywords and patterns) in the user's replies. So the way the chatbot proceeds depends on what has been said before. Furthermore, we want to find out more about the user so that we can determine whether there are features about the user such as personality traits that suggest that the user would put forward certain arguments or reject certain arguments. The dialogues acquired by the chatbot will be useful data for designing interventions in healthcare. In a sense, the chatbot will be an automated researcher for carrying out surveys with users. In particular, we want to identify arguments that particular kinds of user might have that we could then incorporate in automated persuasion systems such as being developed in the EPSRC project Framework for Computational Persuasion.