Development and evaluation of a novel virtual agent-based app for patients with colorectal cancer: A mixed methods study

Abstract

Background and aim: Information support is an integral part of cancer care, but its provision can be problematic in busy health settings. The aim of this project was to develop and evaluate a health app to facilitate the provision of information support in newly diagnosed patients with colorectal cancer (CRC). Instead of delivering information using text, three animated embodied virtual agents (VAs) were deployed. The VAs were formulated after patients’ treating clinicians (male oncologist, female nurse and female pharmacist) to explore the role of familiarity, which has not been addressed in previous research.

Study methods: A multi-stage development process was followed for the app, which was provided to the study participants before the beginning of their treatment. A convergent parallel mixed methods design involving pre- and post-exposure questionnaires (adapted versions of the Toronto Information Needs Questionnaire and the System Usability Scale), app usage data and semi-structured interviews was deployed to evaluate the intervention.

Results and discussion: The app was acceptable by the end users and had a good degree of usability (mean System Usability Scale score=73.89). The information content was appropriate and met patients’ demands to a moderate extent; this was because patients utilised other information sources (e.g., printed material) to address their needs. Incorporating supportive functions such as a medicinal calendar in addition to the information content emerged as an important aspect.
The inclusion of VAs was deemed to be appropriate. The VAs fostered a sense of presence, added trustworthiness to the information content and were perceived as more interactive than reading text. Having a VA representing a familiar clinician was favoured by most users. The vast majority of patients perceived the VAs as cartoon figures and suggested that they should be improved to look realistic in order to give the impression of having an exchange with a real person. Natural voices were preferred over synthetic speech.

Conclusion: VA-based mHealth interventions are an acceptable way of supporting patients with CRC. Appropriate consideration should be given to the requirements of the intended user audience to design acceptable interventions that reflect their needs.