Investigating the Automation of Self-Help Therapies by means of Visual Tooling

Abstract

There are many self-help therapies around, some are written in the form of books while others are internet-based. These forms share similarities but they also have their differences. One of these differences is the transfer of informa- tion. Where books only convey information one-way, internet-based therapy allows for interactivity between therapist and client. An e-coach is an exam- ple of a mobile agent that coaches their clients (coachees) through a process of behaviour change. These e-coaches mostly focus on a single therapy domain and are developed specifically for that. However, the creation of these e-coaches requires extensive technical knowledge and programming skills. Throwing up a barrier for writers of self-help books to develop their own e-coach. We propose a framework for an authoring tool (CoachLab), that allows au- thors of self-help treatments to develop e-coaches without technical knowledge. CoachLab is unique in its underlying generic framework that allows authors to generate conversation-based e-coaches for all therapy domains by only provid- ing therapy specific variables. For this research we focused on the development and generation of conversations. These conversations were created by analysing existing self-help books and e-coaches. Since CoachLab is based on conversations, it is important for authors to convey their message properly. Even though CoachLab enables authors to gen- erate these conversations based on specified variables, authors should be able to adapt them to their own insights. This should again be possible without any programming knowledge. The built-in dialogue editor allows authors to adapt existing and create fully new conversations. To conclude, we created a gener- alising framework for creating conversation-based e-coaches based on existing therapies and processes, along with an editing tool that removes the technical threshold for authors designing those conversations.